PA450/4-90-007G
United States
Environmental
Protection
Agency
AIR
Office of Air Quality
Planning and Standards
Research Triangle Park, NC 27711
U50/4-90-607G
June 1992
PA USER'S GUIDE FOR THE
URBAN AIRSHED MODEL
Volume VII: User's Manual for the
Performance Evaluation System
-------
EPA-450/4-90-007G
USER'S GUIDE FOR THE
URBAN AIRSHED MODEL
Volume VII:
User's Manual for the Performance Evaluation System
OFFICE OF AIR QUALITY PLANNING AND STANDARDS
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711
JUNE 1992
-------
NOTICE
The information in this document has been funded wholly or in part by the U.S. Environ-
mental Protection Agency (EPA) under contract 68-01-7365 to Computer Sciences Corporation.
It has been subjected to the Agency's review, and it has been approved for publication as an EPA
document. Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
-------
PREFACE
This user's guide for the Urban Airshed Model (UAM) is divided into seven volumes as
follows:
Volume I User's Manual for UAM(CB-IV)
Volume II User's Manual for the UAM(CB-IV) Modeling System
(Preprocessors)
Volume III User's Manual for the Diagnostic Wind Model
Volume IV User's Manual for the Emissions Processor System
Volume V Description and Operation of the ROM-UAM Interface Program
System
Volume VI User's Manual for the Postprocessing System
Volume VII User's Manual for the Performance Evaluation System
Volume I provides historical background on the model and describes in general the
scientific basis for the model. For those users that already possess a UAM modeling database
or have prepared inputs without the use of the standard UAM preprocessors, this volume should
serve as a self-sufficient guide to running the model.
Volume II describes the file formats and software for each of the standard UAM
preprocessors that are part of the UAM modeling system. Included in this volume is an example
problem that illustrates how inputs were created from measurement data for an application of the
UAM in Atlanta.
Volume III is the user's manual for the Diagnostic Wind Model (DWM). This model is
a stand-alone interpolative wind model that uses surface- and upper-level wind observations at
selected sites within the modeling domain of interest to provide hourly, gridded, three-
dimensional estimates of winds using objective techniques. It provides one means of formulating
wind fields to the UAM.
-------
Volume IV describes in detail the Emissions Processor System (EPS). This software
package is used to process anthropogenic area and point source emissions for the UAM. An
appendix to this volume describes the Biogenic Emissions Inventory System (BEIS), which can
be used to generate gridded, speciated biogenic emissions. Software for merging the
anthropogenic area, mobile, and biogenic emissions files into UAM input.format is also described
in this volume.
Volume V describes the ROM-UAM interface program system, a software package that
can be used to generate UAM input files from inputs and outputs provided by the EPA Regional
Oxidant Model (ROM).
Volume VI is the user's manual for the Postprocessing System (PPS). The UAM
Postprocessing System (UAMPPS) is a data display and analysis tool for evaluating emission
control strategies. After executing the UAM system, the results from one or more model runs
may be graphically compared. The UAMPPS creates time-series plots, tile maps, gridded value
maps, bar charts, box plots, scatter plots, and quantile plots. Through a series of menus, the user
has a wide range of flexibility in scaling and labelling the graphs.
Volume VII is the user's manual for the Performance Evaluation System (PES). The
UAM Performance Evaluation System (UAMPES) is a data display and analysis tool, which
allows statistical and graphical comparisons between UAM predictions and actual observations
contained in the Aerometric Information Retrieval System. These comparisons allow users to
evaluate UAM performance. The variety of graphs and the system of menus used to create the
graphs are similar to the UAMPPS. Experience acquired in using the UAMPPS should facilitate
the use of the UAMPES.
IV
-------
CONTENTS
Preface iii
Figures vii
Tables ix
1. INTRODUCTION 1
1.1 Background 1
1.2 The UAMPES 2
1.3 Organization of This Manual 2
1.4 For More Information 2
1.5 Conventions Used in This Manual 3
2. OVERVIEW 5
2.1 Introduction 5
2.2 Statistical Performance Measures 6
2.3 Graphical Performance Procedures 8
2.4 Cell-Aggregation Methods 20
3. PROCEDURES 23
3.1 Authorization 23
3.2 Equipment 23
3.3 Getting Started 24
3.4 Using Menus 25
4. TUTORIAL 31
4.1 Starting the UAMPES 31
4.2 UAMPES Main Menu 32
4.3 Selecting a Graphic 33
4.4 Generating a Bar Chart 34
4.5 Saving the Graphic in a Graphics Catalog 47
-------
CONTENTS (concluded)
5. MENU REFERENCE 49
5.1 Menu Outline 49
5.2 UAMPES Main Menu 52
5.3 Graphic Selection Menu 53
5.4 Method/Type Selection Menus 55
5.5 Data File Selection Menu and Print File Selection Window 60
5.6 Monitor Selection Menu 63
5.7 Attribute Selection Menus 65
5.8 Graphic Title/Description Specification Menu 76
5.9 Axis Scaling Menu 78
5.10 Termination Selection Menu 80
5.11 Graphics Catalog Generation Menu 81
6. FILES 83
6.1 Format 83
6.2 SAS Graphics Catalog 83
6.3 Tables 84
REFERENCES 95
APPENDIX A. Data Extraction A-l
APPENDIX B. Performance Measure Formulations B-l
APPENDIX C. Algorithms for Cell-Aggregation and Fractional Bias C-l
C.I Distance-Weighted Algorithm C-l
C.2 Four-Cell Algorithm C-2
C.3 Bilinear Interpolation Algorithm C-4
C.4 Fractional Bias Algorithm C-6
APPENDIX D. Graphics Catalog D-l
-------
FIGURES
Number Page
2-1 Summary statistics table 7
2-2 Shaded tile map (MAPI) of New York domain with cities and monitors
annotated 11
2-3 Time-series line graph (TIME1) 13
2-4 Bar chart (BAR 1) 14
2-5 Scatter plot (SCAT1) 16
2-6 Quantile plot (QQ1) 17
2-7 Bias plot (BIAS1) 19
2-8 Square of nine cells 21
3-1 UAMPES Main Menu 24
3-2 Example UAMPES menu 25
3-3 Example UAMPES menu without instructions 27
3-4 Example UAMPES menu with error message 28
3-5 Example UAMPES menu with a scrollable list 29
4-1 Invoking the UAMPES from the TSO 31
4-2 UAMPES Main Menu 32
4-3 Graphic Selection Menu 33
4-4 Bar Chart Method/Type Selection Menu 35
4-5 Data File Selection Menu 36
4-6 Data File Selection Menu with Print File Selection window 38
4-7 Monitors Located Within Domain Menu 39
4-8 Monitors Located Within Domain Menu with modified labels 40
4-9 Bar Chart Attribute Selection Menu 41
4-10 Bar Chart Attribute Selection Menu with a Select Data window 42
4-11 Graphic Title/Description Specification Menu 43
4-12 Bar Chart Axis Scaling Menu 44
4-13 Bar chart (BAR1) from the tutorial 45
4-14 Termination Selection Menu 46
4-15 Graphics Catalog Generation Menu 47
4-16 PROC GREPLAY window 48
5-1 The UAMPES menu structure 50
-------
FIGURES (concluded)
Number Page
5-2 UAMPES Main Menu 52
5-3 Graphic Selection Menu 53
5-4 Spatial Graphics Annotation Selection Menu 55
5-5 Time-Series Method/Type Selection Menu 58
5-6 Data File Selection Menu 60
5-7 Data File Selection Menu with Print File Selection window 62
5-8 Monitors Located Within Domain Menu 63
5-9 Spatial Graphics Pattern Selection Menu 66
5-10 Statistical Composition of Data Menu 66
5-11 Cities Located Within Domain Menu 68
5-12 Time-Series Attribute Selection Menu 69
5-13 Bar Chart Attribute Selection Menu 71
5-14 Scatter Plot Attribute Selection Menu 72
5-15 Quantile Plot Attribute Selection Menu 74
5-16 Bias Plot Attribute Selection Menu 75
5-17 Graphic Title/Description Specification Menu 76
5-18 Bar Chart Axis Scaling Menu 78'
5-19 Termination Selection Menu 80
5-20 Graphics Catalog Generation Menu 81
6-1 TIME2 graph and table 85
6-2 Summary statistics table 87
6-3 Spatial graphic tables 89
6-4 Quantile plot table 90
6-5 Bias plot tables 91
6-6 Comparison of (a) TIM El and (b) SCAT1 tables 93
6-7 Comparison of (a) TIME1 and (b) BAR1 tables 94
C-l The square of four cells C-3
C-2 Bilinear interpolation C-5
C-3 Source of statistics for fractional biases C-7
Vlll
-------
TABLES
Number Page
1-1 Conventions used in this manual 3
2-1 Graphics supported by the Performance Evaluation System 9
2-2 The six cell-aggregation methods 21
3-1 PF-key functions for UAMPES menus 28
IX
-------
1. INTRODUCTION
This manual describes the procedures for using the Performance Evaluation System of the
Urban Airshed Model (UAM). This section summarizes the purpose and features of the
Performance Evaluation System and outlines the organization of the rest of the manual.
1.1 BACKGROUND
The 1990 Clean Air Act Amendments require all States with ozone nonattainment areas to
submit a State Implementation Plan (SIP), which will explain the emissions control methods
that each State plans to implement to achieve attainment. A concurrent requirement for some
areas is to deliver a UAM modeling demonstration in support of the SIP.
The U.S. Environmental Protection Agency (EPA) issued guidance (EPA, 1991a) that
provides recommendations and general procedures for exercising the UAM in regulatory
applications. In particular, Chapter 5 of the UAM guideline document describes a set of
performance statistics and graphical procedures that are useful to assess the accuracy of UAM
in predicting ozone levels in specific applications areas.
The UAM Performance Evaluation System (UAMPES) was developed to provide assistance in
implementation of the performance measures described in Chapter 5 of the UAM guideline
document. The UAMPES enables users to compute and tabulate the seven specific statistical
measures of model performance recommended in the UAM guideline document. In addition,
the system enables users to graphically display comparisons between model predictions and
measured concentrations using a variety of formats indicated in the UAM guideline document.
The UAMPES is designed to use two data sets.
A data set of predicted concentrations is generated by the Extract module, which is
common to both this system and the UAM Postprocessing System (UAMPPS).
This data set contains data for a particular pollutant, geographical area, layer
height, and time frame. A 3-dimensional Cartesian grid is laid over the
geographical area. The data set contains predicted pollutant concentrations for
each grid cell in the layer of cells nearest the earth's surface.
A data set of observed concentrations is derived from the Aerometric Information
Retrieval System (AIRS) AMP350 workfile, and contains observations for specific
1 1. INTRODUCTION
-------
monitoring stations. The monitoring stations are referred to as monitors in this
manual and in the graphs and tables generated by the UAMPES.
By specifying the same geographical area for both data sets, users can make comparisons
between the observed concentration levels recorded by the monitors and the concentration
levels predicted by the UAM. These comparisons can then be used to evaluate model
performance.
1.2 THE UAMPES
The UAMPES is available on the IBM computer systems at the EPA National Computer
Center (NCC) and may be used by anyone with access to those systems. The UAMPES uses
menus to elicit necessary information for producing the graphics. The menus enable even
novice users to execute the UAMPES.
1.3 ORGANIZATION OF THIS MANUAL
The following sections of this manual provide the information necessary for using the
UAMPES.
Section 2 describes the features and characteristics of the UAMPES.
Section 3 describes the equipment you need and how to start the UAMPES.
Section 4 is a tutorial that illustrates a typical session.
Section 5 describes each menu and how to use it.
Section 6 describes the ways in which the output may be stored and/or printed.
1.4 FOR MORE INFORMATION
This manual provides complete instructions for using the UAMPES, but it does not attempt to
cover other topics. For additional information contact the sources listed below.
More information about the data extraction is given in Appendix A; this appendix is identical
in content to Section 4.1 of the user's manual for the UAM postprocessing system (EPA,
1990a).
For information about the AIRS AMP350 Work file, see the Aerometric Information Retrieval
System (AIRS) user's guide, Volumes IV and V (EPA, 1991b). These user's guides are
available from the meteorology/modeling contact for your EPA Region, or from the National
Technical Information Services at:
1, INTRODUCTION 2
-------
National Technical Information Services
5285 Port Royal Rd.
Springfield, VA 22161
Telephone: (703) 487-4650
For more information about using the IBM computer system at the NCC, or for help with
IBM computer-related problems, contact the EPA User Support Department. Specialists are
on duty weekdays 8:00 a.m. through 7:00 p.m. eastern time to help solve problems and to
answer questions. The mailing address and telephone numbers are:
User Support Department, MD-34
EPA National Computer Center
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Telephone: (919) 541-7862 or (800) 334-2405
1.5 CONVENTIONS USED IN THIS MANUAL
The conventions listed in Table 1-1 are used in this manual.
TABLE 1-1. CONVENTIONS USED IN THIS MANUAL
Convention Description
UAMPES Main Menu Italics indicate names of menus
Select Boldface lowercase words indicate a
menu field
< > characters indicate a terminal key.
1. INTRODUCTION
-------
-------
2. OVERVIEW
The UAMPES combines predicted data generated by the UAM and observed data from
monitor observations to generate statistical performance measures and supporting graphics
that can be used to evaluate the accuracy of the model. This section discusses the statistical
performance measures, the cell-aggregation methods, and the graphics that are available in
the UAMPES.
2.1 INTRODUCTION
The UAMPES provides an automated method for calculating and displaying the performance
statistics and graphic comparisons prescribed in the UAM guideline document (EPA, 1991a).
Section 2.2 briefly discusses the seven statistical measures of model performance that are
computed and tabulated by the UAMPES. Section 2.3 presents an overview of (a) the
graphical components of the UAMPES and (b) the link that exists between the statistical
performance measures and the graphical displays. Finally, Section 2.4 discusses methods for
obtaining various spatial aggregates of model predictions used in the observed/predicted
comparison process.
2. OVERVIEW
-------
2.2 STATISTICAL PERFORMANCE MEASURES
The UAM guideline document (EPA, 199la) prescribes seven statistical measures of model
performance, which are to be applied in evaluating the performance of the UAM in a
particular application. The UAMPES allows the user to extract model predictions along with
appropriate ambient measurements of ozone for statistically based comparisons.
Figure 2-1 is an example output from the UAMPES, in tabular form, that summarizes the
seven statistical performance measures. The summary statistics table is determined by: the
monitors selected, the hours selected, and the aggregation method selected. This table is
generated by the UAMPES every time the user selects a graph for which an aggregation
method is specified (i.e., all graphics categories except the spatial graphics category). The
spatial graphics category does not display cell aggregation and there is no opportunity for the
user to specify a cell-aggregation method. The summary statistics are not affected by the
"type" option. For example, the time-series plot, scatter plot, and quantile plot categories
have three type options. The third option in all three categories generates a graph based on
domain-wide maximum predicted values without regard to any aggregation method. The
summary statistics table, which accompanies these graphs, uses predicted data values that are
based on the aggregation method specified in Menu 2.1.x, but the graph displays domain-wide
predictions. The aggregation method selected is the basis for the summary statistics. The
summary statistics use the formulations prescribed in the UAM guideline document (EPA,
1991a).
As shown in Figure 2-1, the output displays notations, units, and the value of each perfor-
mance measure. For each statistic, a value near zero is desirable. The reader is referred to
Appendix B of this manual [which is identical to Appendix C of the UAM guideline
document (EPA, 199la)] for further details on notation and formulas and for explanations of
the statistical terminology. Note that of the seven statistical performance measures, only
fractional bias is appropriately displayed by the UAMPES in a graphical format (see discus-
sion below).
2, OVERVIEW
-------
Summary Statistics of UAM Recommended Performance Measuras
for MEW YORK, 11JUL88, hours 1 through 24
using BILINEAR method and 9 monitors
Parf. Measure
Description
Notation
Unit*
Value
Unpaired highest
prediction accuracy (1)
32.65
Normalized
bias test (1)
-0.91
Gross error for all
pairs > 60 ppb
Ed*
0.26
Average station peak
prediction accuracy
28.17
Bias of all
pairs > 60 ppb (1)
D60
PPb
10.80
Bias of all
station peaks (1) Dpeak
Fractional Bias for Peak Concentration
Peak Monitor/Station Values
ppb
24.57
Fractional bias
of means (1)
Fm
0.32
Fractional bias of
standard deviation (1)
0.26
Peak Hourly Values
Fractional bias
of means (1)
Fm
0.06
Fractional bias of
standard deviation (1)
Fs
Figure 2-1. Summary statistics table.
0.63
(1) Note: For these measures,
a positive value indicates UNDERprediction
and a negative value indicates OVERpredictlon.
2. OVERVIEW
-------
2.3 GRAPHICAL PERFORMANCE PROCEDURES
The UAMPES also enables users to create the graphical displays that directly compare
observed and predicted concentrations. Table 2-1 provides a brief summary of the graphics
categories, including: spatial graphics (maps) of observed and predicted values, time-series
plots, bar charts, scatter plots, quantile plots, and fractional bias plots. Each of the graphic
displays is described in more detail in the UAM guideline document and in Sections 3 and 4
of this document.
Each graphics category has two to four types (see Table 2-1), which produce similar graphs
but display different aspects of the data. For example, for the spatial graphics category, four
types are available. Each observed (monitor location) and predicted (grid cell location)
concentration displayed on the map may represent (1) the average of the concentrations within
the time period, (2) the maximum concentration occurring within the time period, (3) the hour
that the maximum concentration occurred, or (4) the concentrations at a given hour. The
"Number of possible plots" column in Table 2-1 highlights the fact that TIME1 and BAR1
options may generate multiple plots from a single request. The "Space and/or Time pairing"
column indicates whether observed and predicted concentration levels* are unpaired, paired in
space, paired in time, or paired in both space and time. The "Number of data values per plot"
column indicates how many data points will appear in each plot.
In the following subsections, we provide a description of the elements that are common to all
of the graphs and a brief discussion of the different graphics with accompanying figures.
Only a single example is illustrated for each graphics category, because the types are virtually
identical in appearance except for appropriate changes in titles and labels. However, all of
the graphics categories are described.
2. OVERVIEW
-------
1
r^
J-/3
C/J
Z
O
H
S
j
^<
^
w
U
1
o
fa
C*
A
CM
ffl
H
CQ
C
SUPPORTE
^PHICS
2
O
f
i
U
CO
H
4>
'ra
i
o
t*H
O
Ul *J
_4> _0
1 i!
o £?
§ -c
§ '§.
4> .,
y «
S-l
oo H
S3
'o "E,
II
S ^
5 ^
i
"3,
2
a
c
"o
U
X
1
U
E
P
<*!
c.
00
o
3
s
Average concen
S
o
w
on
.2
I
C/3 CX2
C C
i 1
"o "o
u O
X X
o o
05 Ct!
O O
&O C/3
.2
'^J
2 r""^
c S
Maximum cona
Hour of maximi
2 S
c
|
"o
U
x
wo
O
4)
S
p
°y
4)
§,
-g
Q
a
Concentration al
<
S
VH £2 U
33 =
o o o
XXX
XXX
CS eN eN
I § .§
00 P P
'2
o
"T"^ ^^^.
2 §
(Each monitor)
(Maximum mon
(Maximum dom;
-H (N en
WWW
S S S
P P P
-p
«
Ui
oa
a
oo
4)
_s
o
o
X
(N
U
g
P
«%i
c-
3
O
Ui
O
1
W
?
CQ
i
U
53
CQ
v>
u<
O
O
X
eN
a.
|
CQ
i
o
X
g
S S£ S£
S 3 3
& 0 0
2 x x
4>
S
P
=^
ll 1
oo H H
^
Ul ^~i *~*^
O -"S TO
(All monitors, h
(Maximum mon
(Maximum dom
U U U
oo oo oo
3
"S-
s
1
g
o
x
Monitors
Hours
o -o
4> CJ
a -a
C- C.
P D
In 'w?
11 monitors, houi
[aximum monito
< 2
i (N
a a
a a
"H
1)
n
3
a
3
o
X
"S
1
D
^^
[aximum domain
g
( t i-H
D rt,
Q g
CC H
ot P
C^
3 _
Maximum moni
Maximum hour)
-H eN
OO OO
*H h <
CQ CQ
o
"a.
2
CQ
-------
2.3.1 Graphics Elements
The graphics produced by the UAMPES generally have five elements in common. These
elements are titles, x and y axes, axis labels, legends, and graphics labels.
Titles are located at the top of the graphic and provide general information about
the data used. The title information includes, but is not limited to, the species of
pollutant and the location and time frame of the study.
The x (horizontal) and y (vertical) axes frame the bottom and left side of the plot
and indicate the scale of the units being plotted.
The axis labels are oriented along their respective axes and describe the units that
are scaled along each axis.
The legend is enclosed in a box located below the ;c-axis label. The legend
provides a key to the symbols used in the plot.
The graphics label is located in the lower right hand corner and provides a quick
reference to tie together the graphics, the tables that accompany them, and their
description in this manual.
2.3.2 Spatial Graphics
Spatial graphics present the grid cells associated with the predicted data overlaid on a map of
the state and county lines of the geographical area. There are four types of spatial graphics,
i.e., four ways of reporting predicted concentration levels: average concentration for a given
time period, maximum concentration for a given time period, hour at which the maximum
occurs during a given time period, and concentration at a given hour. The first three types
(MAPI, MAP2, and MAPS) consider concentrations over a time period, or range of hours,
specified in the Data File Selection Menu. If the user specifies a range of 24 hours, the first
type of spatial graphic averages the 24 values for each grid cell and reports the result.
Similarly, the second type locates and reports the maximum value over 24 hours for each grid
cell. The third type reports the hour at which that maximum occurred, recognizing that the
maximum may occur at different hours in different grid cells. The fourth type (MAP4)
reports cell concentrations for a single hour specified by the user. Observed data are reported
in the same manner, except monitor locations are used instead of grid cell locations.
There are two different display options for each of the four spatial graphic types. In the
shaded tile map, in Figure 2-2, each grid cell (or tile) can be a different color or shade
depending on the predicted concentration recorded in the extracted data set. In the gridded
value map (not illustrated), the predicted numerical concentration value is plotted at each grid
cell location. Observed concentrations are plotted at the location of the monitors.
2, OVERVIEW 10
-------
AVERAGE 03 VALUES FOR LEVEL 1 FOR NEW YORK
11JUL88 FOR HOURS 1 THRU 24
CONCENTRATION (PPB) < 60 < 70 » < 80 * < 90 < 100 Bi > 100
Figure 2-2. Shaded tile map (MAPI) of New York domain with cities and monitors
annotated.
There are four general ways to vary the two different display options for spatial graphics:
(1) You may select either a color or monochrome plot.
(2) You may annotate the grid divisions with either column and row numbers derived
from the UTM coordinate system or latitude and longitude values. You may also
choose to have no grid division annotation.
(3) You may choose to annotate city names and locations. If you decide to annotate
cities, you may choose all the cities in the SAS map database or you may select a
subset of those cities to annotate. If you have already selected a subset of cities you
may also choose to simply repeat the previously selected subset.
(4) The concentrations at the monitors listed in the AIRS AMP350 workfile may also be
plotted on the map. The options for the monitors are similar to the options for the
cities. You may choose to plot all or only a subset of the monitors.
11
2. OVERVIEW
-------
2.3.3 Time-Series Line Graphs
The time-series line graph has three types:
1. observed and predicted concentrations, by monitor (multiple plots, one for each
monitor location);
2. maximum observed and predicted concentrations across all monitor locations,
and;
3. domain maximum observed and predicted concentrations.
For types 1 and 2, the predicted concentrations are derived by the user-specified, cell-
aggregation method discussed in Section 2.4. For type 3, the predicted concentrations are the
highest predicted concentrations occurring throughout the domain (i.e., cell aggregation does
not occur). All three types plot two lines: one for observed concentrations and one for
predicted concentrations, over the hours requested by the user. The' hours requested must be
part of the UAM extracted and AMP350 data sets.
The first type (TIME1), illustrated in Figure 2-3, plots values obtained by selecting the
observed and predicted concentrations (paired in space and time). Observed values are com-
pared to the predicted values obtained by the selected cell-aggregation method (as discussed
in Section 2.4). TIME1 produces a separate graph for each monitor in the AMP350 workfile
unless you specify fewer monitors. The procedure for specifying fewer monitors is discussed
in Sections 5.4.2 and 5.6 of this manual.
The second type (TIME2) selects the maximum observed and predicted concentrations (paired
in time, not space) over all monitors for each hour. TIME2 also uses predicted values
generated by one of the six cell-aggregation methods. TIME2 creates a single graph with two
lines: one for maximum observed concentrations and one for maximum predicted concentra-
tions over the hours requested.
The third type (TIME3) selects the maximum observed and predicted concentrations (paired
in time, not space) over all monitors for each hour, and predicted values from all of the grid
cells in the domain (i.e., cell aggregation does not occur). Again, separate lines are plotted
for maximum observed concentrations and maximum predicted concentrations.
12 2. OVERVIEW
-------
03 VALUES FOR NEW YORK, 11JUL88
HOURS 1 THRU 24
MONITOR I.D.=340170006
200
ISO
180
170-
160
ISO-
140-
' ISO-
. 120-
110-
100-
90-
80
70-
60-
5O
40-
30-
20-
10-
0-
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
HOUR
METHOD
I OBSERVED
WT 4 CELL
TIME I
Figure 2-3. Time-series line graph (TIME1).
13
2. OVERVIEW
-------
2.3.4 Bar Charts
There are two possible types of bar charts: observed and predicted concentrations at each
hour, and maximum observed and predicted concentrations occurring within a user-specified
time period. Both types display a pair of bars (one for observed concentrations and one for
predicted concentrations) for each monitor location in the domain. The height of each bar is
proportional to the concentration level. The percent difference between the two bars [i.e.,
{(observed - predicted) / observed} x 100] is displayed numerically above the bar for the
predicted concentration level. Predicted concentrations are generated by the user-specified,
cell-aggregation method.
The first type (BAR1), illustrated in Figure 2-4, displays observed and predicted concen-
trations (paired in space and time). Each graph displays a pair of bars (representing observed
and predicted concentrations) for each monitor selected by the user. Note that a separate
graph is produced for each hour requested. To prevent multiple graphs, specify a single hour
in the Data File Selection menu.
The second type (BAR2) displays the maximum observed and predicted concentrations
(paired in space, not time) for each observed and predicted pair over all hours requested.
Again, multiple pairs of bars are shown on the graph, but only one graph is produced for this
graph type. A monitor label is used to identify the spatial location of each pair of concentra-
tions.
03 VALUES FOR NEW YORK. 11JUL88
(WITH PERCENT DIFFERENCE ANNOTATED)
HOUR=16
0900]
09007
34017 mi 36081 361(13 36119
METHOD
OBSERVED
WT 9 CELL
SARI
Figure 2-4. Bar chart (BAR1).
2. OVERVIEW
14
-------
2.3.5 Scatter Plots
There are three types of scatter plots: observed and predicted, maximum observed and
predicted, and domain maximum observed and predicted. For each type, predicted concen-
trations are plotted on the ordinate [vertical (y) axis] against observed concentrations on the
abscissa [horizontal (x) axis]. Figure 2-5 presents an example scatter plot.
Each scatter plot type displays five reference lines to aid in gauging the fit between observed
and predicted concentrations. The x:y (observed:predicted) ratios of the reference lines are
1:2, 1:1.3, 1:1, 1.3:1, and 2:1. The 1:1 line is a solid line; the others are dashed. The 1:1
line indicates that the predicted value is exactly equal to the observed value. This line is
annotated as 0%, i.e., no difference between the observed and predicted values. The 1:2 line
indicates that the predicted value is two times or 100% larger than the observed value and is
annotated as -100%. The 1:1.3 line indicates that the predicted value is 1.3 times or 30%
larger than the observed value and is annotated as -30%. Conversely, the 1.3:1 line indicates
that the predicted value is 30% smaller than the observed value, and the 2:1 line indicates that
the predicted value is 100% smaller than the observed value. The x:y ratios and the corre-
sponding location of the reference lines can be changed (see Section 5.7.5).
In the first type (SCAT1), illustrated in Figure 2-5, observed concentrations are plotted
against predicted concentrations generated by the selected cell-aggregation method (i.e., paired
in space and time). The number of points on the graph are equal to the number of hours
requested times the number of monitors in the requested domain.
The second type (SCAT2, not illustrated) selects the maximum observed and predicted
concentrations of all monitors for a selected cell-aggregation method and for the requested
hours (paired in time, not space). There is one point for each hour requested.
The third type (SCAT3, also not illustrated) is similar to the second type (SCAT2), except the
maximum predicted values come from all of the grid cells in the domain of the extracted data
set. Again, there is one point for each hour requested.
2, OVERVIEW 15
-------
03 VALUES FOR NEW YORK, 11JUL88
9 MONITORS FOR HOURS 1 THRU 24
REFERENCE LINES AS INDICATED
200-
175-
. 150-
125-
50 :
25-
0 25 50 75 100 125 150 175 200 225
OBSERVED CONCENTRATION (PPB)
SCAT1
Figure 2-5. Scatter plot (SCAT 1).
2.3.6 Quantile Plots
Quantile plots are scatter plots using ranked data with the deciles annotated. The data are
identical to the data for scatter plots discussed above and the types (QQ1, QQ2, and QQ3)
correspond to the scatter plots types. In QQ1, predicted concentrations generated by the
selected cell-aggregation method are plotted against observed concentrations. QQ2 selects the
maximum observed and predicted concentrations of all monitors for a selected aggregating
method and for the requested hours. QQ3 is similar to QQ2 except that the maximum
predicted values come from all of the grid cells in the domain of the extracted data set.
For each of these three types of quantile plots, the data are first scanned for any missing
values in either the observed or predicted concentrations. If a missing value is detected, the
whole observation (including both observed and predicted concentrations) is discarded. Thus,
the data set will have the same the number of observed concentration values as predicted
concentration values.
2. OVERVIEW
16
-------
Observed and predicted concentrations are then independently ranked from highest to lowest
values. The highest observed concentration may occur at hour 16 in monitor A. The highest
predicted concentration may occur at hour 9 in monitor E. Whether or not the highest
observed and predicted concentrations occur at the same time or in the same place is of no
consequence in this plot. Figure 2-6 shows an example quantile plot, for the first type, QQ1
(observed and predicted from aggregate).
As is the case with the scatter plots, the predicted values are plotted on the ordinate and
Observed values are plotted on the abscissa. The label for the ordinate indicates the cell-
aggregation method used. Minimum, maximum, and decile values are annotated and plotted
as filled circles. Intermediate values are plotted as empty circles.
03 VALUES FOR NEW YORK, 11JUL88
5 MONITORS FOR HOURS 8 THRU 20
OEQIES AND REFERENCE UNES AS INDICATED
225
200
f 178
Q.
8 160
B 125-
I
u
u
$1 75-
25
0
0° / ' / x '
V '.' / '
i" " i" T '" i" " I TTTTT " " i'"' i" " i"' n
0 25 50 75 100 125 150 175 200 225
RANKED OBSERVED CONCENTRATION (PPB)
001
Figure 2-6. Quantile plot (QQ1).
17
2. OVERVIEW
-------
2.3.7 Bias Plots
Bias plots in the UAMPES are specifically designed to plot fractional biases. Fractional
biases generally provide a way of comparing two values and indicating how closely one
matches the other. In the context of the UAMPES, positive values of the fractional bias
indicate that the observed concentration statistic (mean or standard deviation) is larger than
the predicted concentration statistic (i.e., underprediction). Negative values indicate that the
observed concentration statistic is smaller than the predicted statistic (i.e., overprediction).
Calculation of fractional bias is described in Appendix C. For a discussion of its use as a
measure of discrepancy, refer to Cox and Tikvart (1990).
The bias plot, illustrated in Figure 2-7, plots the fractional bias of the standard deviation on
the vertical (y) axis against the fractional bias of the mean on the horizontal (x) axis. The
plot displays cross hairs corresponding to the centers of the x and y axes and a pair of boxes
or squares around the center of the plot. Values on both axes range from negative 2.0 to
positive 2.0, with a cross hair at zero. The inner box indicates the point at which the
observed concentration statistic is 30% larger or smaller than the predicted concentration
statistic. The outer box indicates the point at which the observed concentration statistic is
two times larger or smaller than the predicted statistic. The size of the boxes is determined
by factor values of 1.3 and 2.0. For reference purposes, fractional bias values of -0.67 and
-0.26 correspond to overprediction of 100% (factor of 2) and 30% (factor of 1.3), respective-
ly. If you wish, these factor values and the corresponding size of the boxes can be changed.
The procedure for changing these values is discussed in Sections 5.7.5 and 5.7.7 of this
manual.
The first type of bias plot (BIAS1) finds the maximum observed and predicted concentration
values (across hours for each monitor requested by the user), and calculates the means and
standard deviations for both observed and predicted concentrations based on the number (AO
of monitors specified by the user. Thus, these data are paired in space, but not time.
The second type (BIAS2) finds the maximum observed and predicted concentration values
(across monitors for each hour requested by the user) and calculates means and standard
deviations based on the number (N) of hours specified by the user. These data are paired in
time, but not space.
Appendix C contains a detailed description of the methods and calculations used for both
types of bias plots.
18 2. OVERVIEW
-------
MAXIMUM 03 VALUES FOR HOURS 1 THRU 24
FOR NEW YORK, 11JUL88 N=9 MONITORS
2.0-
t.5-
1.0-
0.5-
0.0-
-0.5-
-1.0-
-1.5-
-2.0-
BOXES OUTLINE FRACT. BIAS FACTORS OF 1.3 AND 2
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0
FRACTIONAL BIAS OF MEANS
METHOD
f BILINEAR I
9AS1
Figure 2-7. Bias plot (BIAS 1).
19
2. OVERVIEW
-------
2.4 CELL-AGGREGATION METHODS
Cell aggregation addresses the problem of comparing estimates derived from the UAM (i.e.,
predicted data) with observed concentration data for monitoring sites within the model domain
(i.e., observed data). UAM grid cells are no smaller than 2x2 km and can be as large as 8 x
8 km. In contrast, the air quality data provided by monitors are for specific point locations,
which may not be representative of a single cell or of area-wide concentrations. Cell
aggregation provides a means of comparing estimates from one or more cells for an area
including or close to a monitor location.
The UAMPES offers six cell-aggregation methods. Table 2-2 lists the methods and the
number of cells that each method aggregates. The methods assume an orientation illustrated
in Figure 2-8. For each monitor in the domain, UAMPES determines which grid cell contains
the monitor. The UAMPES selects that cell and the eight cells that surround it (i.e., a square
of nine cells, with the monitor located in the center cell of the square). In the discussion that
follows, each cell is numbered according to the system shown in Figure 2-8. That is, the
cells are numbered 1 through 9 (left to right, from bottom to top). The bottom row cells are
numbered 1, 2, and 3; the middle row cells, 4-6; and the top row cells, 7-9. The monitor (M)
will always be in the center cell (cell 5).
2, OVERVIEW 20
-------
TABLE 2-2. THE SIX CELL-AGGREGATION METHODS
Cell-aggregation method
Number of cells
Same cell
Nine-cell closest value
Distance-weighted nine cell
Four-cell closest value
Distance-weighted four cell
Bilinear interpolation (four cell)
1
9
9
4
4
4
Figure 2-8. Square of nine cells.
(M represents the monitor location.)
21
2. OVERVIEW
-------
The following paragraphs describe the six cell-aggregation methods.
Same Cell The same-cell method is not an aggregation method in the strictest
sense; it only considers a single cell. However, the same-cell method may be the
most obvious comparison to make, because it selects the predicted value for the
cell in which the monitor is actually located (i.e., cell 5 in Figure 2-8). The
predicted value is recorded without weighting and without any calculations other
than those performed by the model.
Nine-Cell Closest Value The nine-cell, closest-value method checks the predicted
concentrations of all nine cells and selects the predicted value that most closely
matches the magnitude of the observed concentration.
Distance-Weighted Nine Cell The distance-weighted, nine-cell method uses an
algorithm to combine all of the predicted concentrations of all nine cells into a
single-weighted average. This average is weighted by the squared inverse of the
distance from the monitor to the center of each cell. Thus, the predicted concen-
trations of the cell centers that are closer to the monitor are given more weight
than the values of the cell centers that are farther from the monitor. The formula
for the distance-weighted algorithm is given in Appendix C.
The three remaining cell-aggregation methods use a common algorithm: the four-cell
algorithm. First, the algorithm selects a square of four cells closest to the monitor. Then, the
individual cell-aggregation methods are applied. Refer to Appendix C for details of the
operation of the four-cell algorithm.
Four-Cell Closest Value The four-cell, closest-value method checks the predicted
concentrations of the four cells selected by the four-cell algorithm and selects the
predicted value that most closely matches the magnitude of the observed concen-
tration. This method is the same as the nine-cell closest value method (described
above), except that four cells are checked instead of nine.
Distance-Weighted Four Cell The distance-weighted, four-cell method considers
the four cells selected by the four-cell algorithm and applies the same algorithm as
the distance-weighted nine-cell method (described above).
Bilinear Interpolation The bilinear interpolation method also arrives at a weighted
average of the four predicted concentrations, but uses bilinear interpolation instead
of the distance-weighting formula. Refer to Appendix C for details on this
algorithm.
2, OVERVIEW 22
-------
3. PROCEDURES
The UAMPES runs on the IBM computer system at the EPA National Computer Center (NCC,
Research Triangle Park, NC). This section explains what authorization you need, what
equipment is required and/or supported, how to start, and how to use system menus.
3.1 AUTHORIZATION
There are no restrictions on the use of the UAMPES on the EPA's IBM computer system at
the NCC. You will need access to the standard facilities of the IBM computer system and the
Time-Sharing Option (TSO). The TSO (the interactive part of the IBM computer operating
system) is the environment in which the UAMPES menus operate. If you are not a registered
user of EPA's IBM computer system, States may get information about registration proce-
dures from the modeling contact for their EPA Region. Others may contact the EPA's NCC
User Support Department identified in Section 1.4 of this manual.
3.2 EQUIPMENT
The UAMPES uses full-screen menus (i.e., the computer system and the terminal exchange
data in screen-sized segments, rather than a line at a time or character-by-character). You
must use a terminal that supports this full-screen protocol, or use equipment that emulates
such a terminal. Also, your terminal must support full-screen graphics to display the graphics
produced by the UAMPES. If your terminal does not support full-screen graphics, you will
need to save the graphics in a catalog (explained in Section 6) and send the graphics from the
catalog to a printer or other hard-copy device.
The UAMPES is designed to support, or produce, graphics on a variety of graphics terminals
and hard-copy devices. If you do not have a full-screen terminal, you can emulate one using
an asynchronous terminal or its equivalent [e.g., a personal computer (PC) equipped with a
modem and appropriate communications software]. The computer system on which the
UAMPES is installed must have a protocol converter (i.e., equipment that converts the
character-by-character communications protocol of an asynchronous terminal to the full-screen
protocol required for the UAMPES menus). For a PC/modem emulation to work acceptably,
the PC communications software must provide a way to make some keys on the PC keyboard
equivalent to the program function (PF) keys on full-screen terminals. Your terminal must be
able to emulate PF keys to use the UAMPES menus.
23 3. PROCEDURES
-------
For more information about terminals or terminal emulations, get in touch with the modeling
contact for your EPA Region or the user support department of the computer system on
which the UAMPES is installed.
3.3 GETTING STARTED
Before you use the UAMPES, you must log on to the TSO of the IBM computer system in
full-screen mode. If you need help doing this, read the Getting Started chapter of the Guide
to NCC Services (EPA, 1990b). This guide is distributed to new users of the NCC when they
register, and current users can request a copy from the EPA User Support Department.
Unfortunately, if you are using an asynchronous terminal or a PC with a modem, the guide is
not very clear about the procedure for connecting to the system in the full-screen emulation
mode. If you need help, get in touch with the modeling contact for your EPA Region, or call
the EPA User Support Department of the NCC (see Section 1 in this manual).
After you have completed the log-on and the TSO has displayed its READY prompt, you are
ready to invoke the UAMPES by entering "UAMPES" (in lowercase or uppercase letters).
Press to transmit the command to the computer and start the UAMPES. There will
be a short pause while the software is loaded and initialized. When the UAMPES Main Menu
emerges (Figure 3-1), the UAMPES is active.
URBAN AIRSHED MODEL PERFORMANCE EUALUATION SYSTEM-
MENU
GRAPHICS
Place cursor on your selection and press the enter key.
Figure 3-1. UAMPES Main Menu.
3. PROCEDURES
24
-------
In Section 3.4, we describe general procedures for using all UAMPES menus. Section 4 uses
an example to illustrate the procedures and menus that you can use to produce a graph.
Section 5 describes the sequence of menus and provides details for each menu.
3.4 USING MENUS
The UAMPES menus make it easy to produce graphics for evaluating the UAM. The
procedure is described in the steps below.
(1) A menu appears on your screen and prompts you for certain information. You
need to supply the requested information and press .
(2) Press . This keystroke will tell the menu program, "I'm done here. Go
on to the next operation." The menu program then checks the information for
errors, omissions, or inconsistencies and reports any problems it finds.
(3) You need to fix the problems,.press , and the next menu will appear.
Steps 1 through 3 are repeated for each menu that is required to define your graphic. After
all the specifications are assembled, the last menu program produces a graphic which will be
displayed on your terminal and/or saved in a graphics catalog.
Figure 3-2 shows an example menu that illustrates many of the features of the UAMPES
menus.
r 2.2
UAMPES: Data File Selection
Please enter the Fully Qualified Data Set Names for the Extracted
UAM (predicted) data set and the AIRS AMP350 (observed) Workfile
to be used in this graphic.
Do not enclose the Data Set Name in quotes.
Extracted Data Set:
AIRS AMP350 Workfile:
Period of Interest: Date Start Hour
End Hour
(DDMMMYY) (1-24)
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
Figure 3-2. Example UAMPES menu.
25 3. PROCEDURES
-------
The following paragraphs describe the menu features and how to use them.
Identification Every UAMPES menu has a number in the top left corner that
uniquely identifies it, and one or two title lines at the top of the screen that
summarize the menu's purpose or function. In Figure 3-2, the identifying number
is 2.2. The method for assigning menu numbers is discussed in Section 5 (Menu
Reference). The title line, Data File Selection, identifies the general function and
purpose.
Instructions Not all menus contain instructions. In the menus that contain
instructions, the lines of text below the menu title explain what you are expected to
do (see Figure 3-2). In the menus that do not contain instructions, the required
action should be self-evident (see Figure 3-3). The specific instructions vary from
one menu to another, but there are two general actions: fill in a blank or select
from a list. In Figure 3-2, the instructions describe filling in blanks. However, in
Figure 3-3, you are expected to select from the list presented and enter the number
of your selection in the accompanying blank.
Selection/Specification Fields and Cursor Movement The middle or bottom of a
menu displays a list of values for you to choose from or has spaces (fields) for you
to enter values. The menu in Figure 3-3 has both. The fields (beside the words
Type:, Aggregate Method:, and Monitor Selection:) will accept values that you
enter to fill in the blank field or replace the initial values. (These fields will be
blank the first time that you run the system, or will contain initial values from the
previous run.) When a menu is first displayed on your terminal, the cursor will
appear at the beginning of the first selection/specification field. Pressing the
key moves the cursor from one field to the next. The cursor will automati-
cally move to the next field if your entry in the previous field completely fills the
space allotted to that field. If your terminal has a or
key, pressing it moves the cursor to the previous field. You can also use the
terminal's cursor movement (arrow) keys to move the cursor to a field, but using
the key is quicker.
Protected Fields The areas of the screen that are skipped over when you press
are protected; you cannot alter them. If you move the cursor with cursor-
-movement keys to a protected part of the screen and try to alter it, the terminal
keyboard locks and you must press the key to regain use of it. If your
keyboard does not have a key, contact the EPA User Support Department
(see Section 1).
Menu titles, instructions, "white space", etc. are protected areas. Sometimes an
area of a menu that looks like a fill-in-the-blank field with an initial value is a
protected field. The menu program determines that only one value is valid for the
3. PROCEDURES 26
-------
-2.1.2
field, inserts that value for you, and makes the field protected. In general, if you
can tab the cursor to a field, it has an initial value that can be changed. If you
cannot tab to it, the field is protected.
PF Keys Table 3-1 (on the next page) lists the standard PF-key functions. The
CANCEL function, or , is available on every menu. It lets you
return to a previous menu or exit from the UAMPES without producing a graph.
The ABORT function, or , is available on every menu except the
UAMPES Main Menu. It is a quick way to abandon a graphics request and return
to the UAMPES Main Menu, but your menu choices are not saved. Some warning
messages may occur if you end a UAMPES session using the ABORT function.
The SCROLL functions listed in Table 3-1 are meaningful only on menus that
display an item list that does not completely fit on one screen.
Error Messages Error messages will appear within a narrow horizontal window in
the middle of the screen if something is wrong. On color terminals, the window
will have a red, flashing border. If you make a mistake when using the UAMPES
menus, this window will appear. If a menu program detects something wrong, it
describes the problem in a one-line message that appears after you press
from your terminal. In Figure 3-4, a message appeared because the value "4" is
entered in the Type field. Only the values "1", "2", or "3" are valid choices in this
field.
UAMPES: TIME-SERIES METHOD/TYPE SELECTION
Type: _ 1. Observed and Predicted (Select Aggregate)
^2. Maximum Observed and Predicted (Select Aggregate)
3. Domain Maximum Observed and Predicted
Aggregation Method: _
Monitor Selection:
PF/l/PF13=Help
Samecell
9 Cell Closest Value
Distance Weighted 9 Cell Average
4 Cell Closest Value
Distance Weighted 4 Cell Average
Bilinear 4 Cell
Use All Monitors
Subset Monitors
Use Previous Subset of Monitors
PF3/PF15=Prev Menu
PF4/PF16=Main Menu
Figure 3-3. Example UAMPES menu without instructions.
27
3. PROCEDURES
-------
TABLE 3-1. PF-KEY FUNCTIONS FOR UAMPES MENUS
PFKey
Command
Assigned Function
PF1 or PF13
HELP
Provides context-sensitive information on how
to proceed or indicates what information is
required.
PF3 or PF15 a
PF4 or PF16 a
PF7 or PF19 b
PF8 or PF20 b
CANCEL
ABORT
BACKWARD
FORWARD
Terminates the current menu without taking
any action and returns to the previous menu.
Terminates the current menu without taking
any action and returns to the UAMPES Main
Menu.
Scrolls toward the beginning of a list of items.
Scrolls toward the end of a list of items.
a These PF keys are available on all menus except the main menu.
b These PF keys are available on menus that contain a list that is too long to fit on one screen.
r-2.1.2
UAMPES: TIME-SERIES METHOD/TYPE SELECTION
Type: _ 1. Observed and Predicted (Select Aggregate)
2. Maximum Observed and Predicted (Select Aggregate)
3. Domain Maximum Observed and Predicted
Press Enter to continue
PLEASE ENTER A 1, 2 OR 3 -OR- PRESS PF13 FOR HELP
Monitor Selection:
PFl/PF13=Help
4. 4 Cell Average
5. Distance Weighted 4 Cell Average
6. Bilinear 4 Cell
I. Use All Monitors
2. Subset Monitors
3. Use Previous Subset of Monitors
PF3/PF15=Prev Menu
PF4/PF16-Main Menu
Figure 3-4. Example UAMPES menu with error message.
3, PROCEDURES
28
-------
Error Correction First, press to remove the error message from the screen.
Fill-in-the-blank fields associated with an error are highlighted. If there are multiple
errors, the error message describes only the first one and highlights the field
associated with that error. You should change that field to eliminate the error.
Then, press . If there is more than one error, another error message will
appear indicating what else needs to be done. In some instances an error condition
involves two or more fields, and you may need to decide which of them is the real
culprit, that is, the one causing the error. Press after each correction.
After no more corrections are needed, press to move on to the next step in
the program.
Scrolling Some menus have a list of items to choose from. The menu shown in
Figure 3-5 below displays a list of monitors which may be selected by entering an
"S" in the Select field. If the entire list does not fit on the screen, you can scroll to
see more items. Imagine that part of the menu is a window through which you see
the list of items. Scrolling forward is like sliding the window down, toward the end
of the list. Scrolling backward is like sliding the window up, toward the beginning
of the list. Scrolling backward at the beginning of a list has no effect, nor does
scrolling forward at the end of a list.
r 2.3
Monitors Located Within Domain
Place an S in the selection column for all monitors desired in this run.
Press PF20 to scroll down the list, PF19 to scroll up.
After all selections have been made, press PF15 to save selections.
Select Bar Label Monitor ID State/County Description
_ 13007 090013007 CT/FAIRFIELD/USCG LIGHTHOUSE, PROSPECT
31003 090031003 CT/HARTFORD/30 REMINGTON ROAD
_ 70007 090070007 CT/MIDDLESEX/CONN. VALLEY HOSP., SHEW
91123 090091123 CT/NEW HAVEN/715 STATE STREET
31001 090131001 CT/TOLLAND/ROUTE 190, SHENIPSIT STATE
_ 70006 340170006 NJ/HUDSON/VETERANS PARK ON NEWARK BAY
_ 10005 340210005 NJ/MERCER/RIDER COLLEGE;LAWRENCE TOWNS
30006 340230006 NJ/MIDDLESEX/RYDERS LANE & LOG CABIN R
10004 360810004 NY/QUEENS/QUEENS COLLEGE, 65-20 KISSEN
30002 361030002 NY/SUFFOLK/EAST FARMINGDALE WATER DIST
92004 361192004 NY/WESTCHESTER/WHITE PLAINS PUMP STATI
Figure 3-5. Example UAMPES menu with a scrollable list.
29 3. PROCEDURES
-------
Field Values Saved and Recalled The first time you use a menu, the fields in
which you can enter values are blank. Underlines indicate the locations and sizes of
blank fields. When you leave a particular menu, the UAMPES saves the values of
all fill-in-the-blank fields, but not the selections you make from a scrollable list.
The next time you use the same menu, the values you entered are displayed as the
initial values. If you want to use the same values again, you do not have to reenter
them. To override any defaults, type over the field and eliminate any extraneous
characters using either the or the or keys.
On-line Help On-line help is available from most screen fields. To access on-line
help, simply place the cursor on the field for which help is desired and press
or . A window with an explanation and additional information on the field
will open on top of the menu. The help window may be one screen or multiple
screens, depending on the field and the complexity of the related information. From
the help window, press or to return to the application screen and
continue your data entry.
Colors on Menus If you are using a terminal that supports color graphics, you may
notice that different types of information are displayed in different colors. The text
describing a given field is displayed in cyan. If the field is protected, the color of
the field will be pink. Fields that require a response from you are displayed in
green; system errors or informational messages are displayed in red. If the system
detects that an invalid value has been entered in a field, that field will be displayed
in white and reverse video. If a monochrome terminal is used, no colors will be
displayed; however, if an error is detected, the terminal will most likely display the
field in reverse video.
3. PROCEDURES 30
-------
4. TUTORIAL
This section illustrates a typical UAMPES session. It shows how the menus will look when
you use the UAMPES, and how the menus respond to common mistakes. You can run this
tutorial on your own terminal by entering the values and commands given in the text.
4.1 STARTING THE UAMPES
Suppose you want to compare concentrations predicted by the UAM with observed concentra-
tions for the New York City metropolitan area. To get started, you will need to have an
extracted data set (described in Appendix A of this manual, from the UAM Postprocessing
System user's guide) and an AIRS AMP350 workfile (described in Volumes IV and V of the
Aerometric Information Retrieval System User's Guide; EPA, 1991b). After completing the
log-on to the EPA's IBM computer system, you are ready to access the UAMPES. Follow
the instructions given in the Getting Started section (Section 3.3), and type "UAMPES" after
the TSO READY prompt. Your terminal screen at this point will look like Figure 4-1.
92/05/19:
92/05/18:
92/05/14:
92/05/14:
92/05/13:
92/05/11:
92/05/11:
92/04/20:
** *
READY
UAMPES
NCC CUSTOMER SUPPORT HOURS EXTENDED
A LIST OF RECENTLY RELEASED EPA MEMOS
COBOL CONVERSION AID 1.6 AVAILABLE
ACCESS JO THE AIRS GRAPHICS SYSTEM
MEMORIAL DAY HOLIDAY SCHEDULE
MVS/ESA V4.2 OPERATING SYSTEM INSTALLED
PRIORITY BATCH JOB PROCESSING ON THE IBM
SUPERCALC 5 . 0 AVAILABLE FOR TESTING
- SEE
- SEE
- SEE
- SEE
- SEE
- SEE
- SEE
SEE
NEWS
NEWS
NEWS
NEWS
NEWS
NEWS
NEWS
NEWS
ALERT3
ALERT 6
ALERT 9
ALERT 10
ALERT 2
ALERT 1
ALERT 8
ALERT5
Figure 4-1. Invoking the UAMPES from the TSO.
31
4. TUTORIAL
-------
4.2 UAMPES MAIN MENU
Press to submit the command and activate the UAMPES. Shortly after you press
, the UAMPES Main Menu appears. As you can see in Figure 4-2, the UAMPES
Main Menu has five possible options from which you may choose. These are: the Extract
module, the Graphics module, the Graphics Catalog module, Help, and Exit. These options
are discussed in detail in Section 5 (Menu Reference).
You may move your cursor from one option to another using the key. To choose an
option, move the cursor to the desired option and press . For this tutorial, move the
cursor to the Graphics module option and press .
URBAN AIRSHED MODEL PERFORMANCE EVALUATION SVSIEM"
MAIN MENU
GRAPHICS CAILG
Place cursor on your selection and press the enter key.
Figure 4-2. UAMPES Main Menu.
4. TUTORIAL
32
-------
4.3 SELECTING A GRAPHIC
Selecting the Graphics module option from the UAMPES Main Menu will bring up the
Graphic Selection Menu illustrated in Figure 4-3. This menu has one list and two selection
fields. The list displays the different graphics available. The selection fields allow you to tell
the system: (1) which type of graphic you wish to produce, and (2) the type of terminal or
hard-copy device that you are using.
The cursor will appear in the first selection field, which is labeled Graphic Selection. Enter
the number that appears beside the graphic you wish to produce. For illustrative purposes,
"3" (Bar Chart) has been entered.
After entering the number in the first selection field, the cursor will automatically move to the
Device field. In the example shown in Figure 4-3, "IBM3179" has been entered in the
Device field. You must enter the appropriate device name for the equipment you are using.
The UAMPES currently supports more than a dozen different devices (terminals, emulators,
etc.). You may view the list of supported devices by typing a "?" in the Device field and
pressing . If you are not sure which name is appropriate for the equipment you are
using, consult the user's manual supplied by the terminal manufacturer or contact the EPA
User Support Department for help (see Section 1).
r 2.0
URBAN AIRSHED MODEL PERFORMANCE EVALUATION SYSTEM (UAMPES)
Graphic Selection
1} Spatial Graphic
2) Time-Series Line Graph
3} Bar Chart
4} Scatter Plot
5) Quantile Plot
6) Bias Plot
Graphic Selection: 3_ Device: IBM3179
Enter "?" on Device field for a list of supported devices.
PFl/PF13=Help PF3/PF15=Exit PF4/PF16=Main Menu
Figure 4-3. Graphic Selection Menu.
33 4. TUTORIAL
-------
After filling in both fields in the Graphic Selection Menu, press . You should then
see the UAMPES menu illustrated in Figure 4-4. However, if the UAMPES detects a
problem with either of the two fields in the Graphic Selection Menu, the menu will remain on
your screen and an error message window containing a one-line description of the error will
appear in the middle of the screen. On color monitors, this error window will be highlighted
with a red, flashing border. On all monitors, there will be a note in the window border
indicating that you first need to press to continue.
After pressing , the error window will disappear and the field containing the error will
be highlighted. Only numbers 1 through 6 are valid in the Graphic Selection field. Other
values will result in an error message. Specifying a device in the Device field that is not
supported will also result in an error message. Note that the entry in the Device field must
exactly match one of the devices in the list of supported devices.
If an error message appears, the cursor will be positioned in the field in which the error
occurs. Simply type over the existing entry to make the correction and press . If
there is more than one error, another error window will appear describing the nature of the
error. Repeat the above sequence, i.e., press key, type over the error to make the
correction, and press .
After you press on the Graphic Selection Menu and the system does not detect any
errors, the system will move on to the next menu.
4.4 GENERATING A BAR CHART
4.4.1 Selecting Type, Method, and Monitor Options
Selecting "3" (Bar Chart) from the UAMPES Graphic Selection Menu will bring up the menu
shown in Figure 4-4. The Bar Chart MethodfType Selection Menu contains three lists and
three corresponding selection fields.
4. TUTORIAL 34
-------
r2.1.3
UAMPES: BAR CHART METHOD/TYPE SELECTION
Type: _!_ 1. Observed and Predicted (Select Aggregate)
2. Maximum Observed and Predicted (Select Aggregate)
Aggregation Method: 3_
Monitor Selection: 2
1. Samecell
2. 9 Cell Closest Value
3. Distance Weighted 9 Cell
4. 4 Cell Closest Value
5. Distance Weighted 4 Cell
6. Bilinear 4 Cell
1. Use All Monitors
2. Subset Monitors
3. Use Previous Subset of Monitors
PFl/PFl3=Help
PF3/PF15=Prev Menu
PF4/PF16=Main Menu
Figure 4-4. Bar Chart Method/Type Selection Menu.
The first list describes the two bar chart types, the second describes the aggregation methods,
and the third describes the monitor options. Refer to Section 2.4 (Cell-Aggregation Methods)
and Appendix C for a complete discussion of aggregation methods. Refer to Section 2.3
(Graphical Performance Procedures) and Section 5.4 (Method/Type Selection Menus) for a
complete discussion of graphic types and monitor options. For the purposes of this tutorial,
simply select the first graphic type, the third aggregation method, and the second monitor
option. These selections are illustrated in Figure 4-4 by the numbers 1,3, and 2 respectively.
The first graphic type, "Observed and Predicted (Select Aggregate)", will produce a bar chart
with a parr of bars for each monitor you choose to include in the analysis. For each pair of
bars, one bar will indicate the concentration level actually observed by that monitor at that
hour. The other bar will indicate the concentration level predicted by the model after
applying the aggregation method that you choose from the second list in this menu.
Choosing "Distance Weighted 9 Cell" in this example will display the weighted-average
concentration of the nine cells closest to the monitor (weighted by the squared inverse of the
distance from the monitor to the center of each grid cell). Selecting the second monitor
option, "Subset monitors", will bring up a monitor selection menu following the Data File
Selection Menu. In this example, the number of monitors chosen only affects the appearance
of the graph. In the quantile plots and bias plots, the number of monitors chosen will also
affect subsequent calculations. Refer to Section 5 (Menu Reference) for a complete discus-
sion of the considerations pertaining to these options.
35
4. TUTORIAL
-------
4.4.2 Data File Selection
The menu shown in Figure 4-4 works the same way as the UAMPES Graphic Selection
Menu. After you have made your selections by entering the appropriate number in each field,
press . The selections illustrated will bring up the Data File Selection Menu shown
in Figure 4-5 below.
At this point you may tell the UAMPES exactly what data to analyze and graph. The Data
File Selection Menu contains five selection fields that must be completed before moving on to
the next menu. There are two long fields for entering data set names. These long fields are
followed by three shorter fields for entering the date, and start and end hours.
The first long field calls for the name of a data set created by the Extract module included in
both this system and the UAMPPS (EPA, 1990a). Enter the fully qualified data set name of
the extracted data set in the space provided. The second long field calls for the name of an
AIRS AMP350 workfile. Enter the fully qualified data set name of the workfile data set in
the space provided. Do not enclose the data set names in quotation marks.
-2.2
UAMPES: Data File Selection
Please enter the Fully Qualified Data Set Names for the Extracted
UAM (predicted) data set and the AIRS AMP350 (observed) Workfile
to be used in this graphic.
Do not enclose the Data Set Name in quotes.
Extracted Data Set: uidacct.UAM.SASD.EXT.O3.XI1JUL88.L1B88
AIRS AMP350 Workfile: uidacct.AMP350
Period of Interest:
Date 11JUL88
(DDMMMYY)
Start Hour 16
End Hour 16
(1-24)
NOTE: This chart type will generate a graphic for each hour
from start hour through end hour. This could be lots of plots!
PFl/PF13=Help
PF3/PF15=Prev Menu
PF4/PF16=Main Menu
Figure 4-5. Data File Selection Menu.
4, TUTORIAL
36
-------
The three short selection fields define the period of interest and consist of the date and
starting and ending hours. The period of interest must be a time frame contained in the two
data sets. The Date should contain seven characters of the form ddmmmyy where dd is the
day, mmm is the month, and yy is the year. For example, if the extracted and workfile data
sets contain data for hours 1 to 24 on August 1, 1991, then you must enter "01AUG91" in the
Date field. You may choose any range of hours between 1 and 24 for the Start and End
Hour fields. The only restriction is that the starting hour must be earlier than the ending
hour.
For this example, notice that the bar chart type, Observed and Predicted (Select Aggregate),
will produce a separate graph for each hour requested. The note at the bottom of the Data
File Selection Menu serves as a reminder of this feature. If you select a Start Hour of " 1"
and End Hour of "24", then 24 graphs will be produced. Producing this number of graphs
may be a waste of your time and/or resources. To produce a single graph, select the same
hour for Start Hour and End Hour (as shown in the example in Figure 4-5).
Only one other graphic type produces multiple graphs: the Observed and Predicted (Select
Aggregate) type of Time-Series Line Graph. All the other types of graphics produce only a
single graph (see Table 2-1).
4.4.3 Print File Selection
For each graph produced by the UAMPES, there will be a corresponding text file produced
that records the exact values or data points plotted in each graph and contains the summary
statistics table (Figure 2-1, Section 2.2). The summary statistics table is generated every time
the user selects a graph for which an aggregation method is specified. The text file will
duplicate the titles and label of the graph it accompanies. The Print File Selection window,
shown in Figure 4-6, allows you to tell the system where to put this text file. The Print File
Selection window will appear on top of the Data File Selection Menu. Refer to Section 5
(Menu Reference) for details of the operation of the Print File Selection window. For the
purposes of this tutorial, enter a three-level file name beginning with your user ID/account
number. The file name should be a name that you are not already using. The UAMPES will
allocate the file for you. In the example shown in Figure 4-6, the name "uidacct.PRINT.-
LISTING" has been entered.
37 4. TUTORIAL
-------
r 2.2
UAMPES: Data File Selection
+-Press ENTER to Continue
Print File Selection
Please enter the Fully Qualified Data Set Name for the text file
that will receive all report output for this session.
Do not enclose the Data Set Name in quotes.
Report File Name: uidacct.PRINT.LISTING
Be sure to record the file name for later use.
PFl/PFl3=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
Figure 4-6. Data File Selection Menu with Print File Selection window.
4.4.4 Selecting a Subset of Monitors
After you have filled in all of the selection fields of the Data File Selection Menu, press
to move on to the Monitors Located Within Domain Menu, as shown in Figure 4-7.
This menu appears because the option to subset monitors was selected in the Bar Chart
Method/Type Selection Menu (Figure 4-4).
Before presenting the list that appears in this menu, the UAMPES compares the two data sets
entered in the Data File Selection Menu, and matches the monitors and their locations that are
in the AIRS AMP350 workfile with the geographical area described by the domain in the
extracted data set. Only the monitors that are located within the domain of the extracted data
set are included in the UAMPES analysis. The monitors that do fit this criteria are then listed
in the Monitors Located Within Domain Menu.
The Monitors Located Within Domain Menu allows you to preview the group of monitors that
will be used in the model performance analysis. You may then select all of the monitors, a
single monitor, or any subset of monitors you wish. The number of monitors selected is an
important consideration in producing bar charts.
The two types of bar charts are the only graphics in the UAMPES that display a monitor
identification (ID) for every monitor in the domain in a single graph. The monitor ID in the
AIRS AMP350 workfile is nine characters long.
4, TUTORIAL 38
-------
r 2.3
Monitors Located Within Domain
Place an S in the selection column for all monitors desired in this run.
Press PF20 to scroll down the list, PF19 to scroll up.
After all selections have been made, press PF15 to save selections.
Select
Bar Label
13007
31003
70007
91123
70006
70006
10004
30002
92004
Monitor ID State/County/Site Address
090013007 CT/FAIRFIELD/USCG LIGHTHOUSE, PROSPECT
090031003 CT/HARTFORD/30 REMINGTON ROAD
090070007 CT/MIDDLESEX/CONN. VALLEY HOSP., SHEW
090091123 CT/NEW HAVEN/715 STATE STREET
340170006 NJ/HUDSON/VETERANS PARK ON NEWARK BAY
340270006 NJ/MIDDLESEX/RYDERS LANE & LOG CABIN R
360810004 NY/QUEENS/QUEENS COLLEGE, 65-20 KISSEN
361030002 NY/SUFFOLK/EAST FARMINGDALE WATER DIST
361192004 NY/WESTCHESTER/WHITE PLAINS PUMP STATI
Figure 4-7.
Monitors Located Within Domain Menu.
If there are more than four or five monitors in the domain, these ID numbers will start to run
into each other. For this reason, the UAMPES will, by default, truncate the monitor ID
numbers to the last five digits of each number. These five digits are placed in the Bar Label
column of this menu.
As long as there are no duplicate entries in the Bar Label column, this system will uniquely
identify each monitor, and as many as 15 monitors can fit on a single graph. However, if you
have any duplicate labels or more than 15 monitors in the requested domain, you will have to
make adjustments in the Select field and/or the Bar Label field.
First, let's deal with the possibility of having more than one monitor with the same value in
the Bar Label field. If there are two or more monitors with the same Bar Label field, the
UAMPES will treat them as multiple observations from a single monitor, add their concentra-
tion levels together, and display the resulting sum. Thus, if levels are running around
100 parts-per-billion (ppb) and there are three monitors with the same Bar Label, the graph
will indicate a monitor with concentration levels around 300 ppb! To prevent this from
happening, be sure that each monitor listed in Figure 4-7 has a unique Bar Label. In
Figure 4-7, note that there are two monitors with the same value ("70006") in the Bar Label
column. Type over the characters in the Bar Label field with any identification that will
work for you (Figure 4-8 shows one possibility). Note: you may enter any combination of
alphabetical or numerical characters. In the example shown in Figure 4-8, the monitors are
now labeled with the first five digits of the Monitor ID, which correspond to the state and
county codes. This strategy will work as long as there is only one monitor in each county.
The entries that appear in the Bar Label field are the ones that will identify each monitor in
the bar chart. Note also in Figure 4-8 that selected monitors are indicated by an "S" in the
Select field.
39
4. TUTORIAL
-------
/d.J - - ' -
Monitors Located Within E)omain
Place an S in the selection column for all monitors desired in this run.
Press PF20 to scroll down the list, PF19 to scroll up.
After all selections have been
Select
S
S
S
S
S
S
S
S
S
Bar Label
09001
09003
09007
09009
34017
34027
36081
36103
36119
Monitor ID
090013007
090031003
090070007
090091123
340170006
340270006
360810004
36103000?
361192004
made, press PF15 to save selections.
State/County/Site Address
CT/FAIRFIELD/USCG LIGHTHOUSE, PROSPECT
CT/ HARTFORD/30 REMINGTON ROAD
CT/MIDDLESEX/CONN. VALLEY HOSP . , SHEW
CT/NEW HAVEN/715 STATE STREET
NJ/ HUDSON/ VETERANS PARK ON NEWARK BAY
NJ/MIDDLESEX/RYDERS LANE & LOG CABIN R
NY /QUEENS /QUEENS COLLEGE, 65-20 KISSEN
NY /SUFFOLK /EAST FARMINGDALE WATER DIST
NY/WESTCHESTER/WHITE PLAINS PUMP STATI
Figure 4-8. Monitors Located Within Domain Menu with modified labels.
Now let's look at the possibility of having more than 15 monitors. The primary limiting
factor on the number of monitors that can be plotted on a single graph is the size of the Bar
Label. If the Bar Label is too large, no graph will be produced. With sufficiently small Bar
Label, more than 80 monitors have been plotted on a single graph in simulation trials. The
size of Bar Label may be reduced by giving each monitor only a two or three character
identifier in the current menu. The size may also be reduced by specifying a smaller size of
print in the Size field of the Graphic Title/Description Specification Menu.
Although more than 15 monitors can be plotted, certain problems arise as the number of
monitors increases. Specifically, the size of the print of the Bar Label may be reduced to the
point of illegibility; percent difference annotations start to overlap and also become illegible;
and the bars themselves begin to resemble lines rather than bars.
To prevent these problems, you may wish to subset the monitors into two or more groups.
For example, if you are considering 20 monitors, you may wish to generate two graphs with
10 monitors on each graph rather than a single graph with all 20 monitors crowded onto it.
To accomplish this, enter an "S" in the Select field that precedes the Bar Label for the first
group of monitors, continue through the UAMPES to produce a graph, and return to this point
to select the next group of monitors.
Only 11 monitors can be listed on the screen at one time, so if there are more than 11
monitors to choose from, you will need to use the or key and the or
key to scroll through the list. Use to move the cursor to the Select field for
4. TUTORIAL 40
-------
each monitor you wish to include in this graph. Type an "S" to indicate that this monitor is
to be included. If you type an "S" and then decide that you do not want this monitor
included in this graph, return the cursor to the Select field (using to delete the "S". After you have determined that each monitor has a unique description
and you have indicated the monitors that you wish to select for this graphic, press the
or key to save the selections and move on to the next menu. The or
key takes the place of in this menu, because the selections to be saved may include
selections that have been scrolled off of the screen.
4.4.5 Selecting Bar Chart Attributes
The next mer^i is the Bar Chan Attribute Selection Menu shown in Figure 4-9. Each of the
six different graphics has a corresponding Attribute Selection Menu that determines the
appearance attributes for that particular graphic.
r2.4.3
UAMPES: Bar Chart Attribute Selection
Bar Hal-
Pattern Color
Observed Values: S RED
Predicted Values: S GRF.EN
Enter "?" for a list of valid entiles
PFl/PF13=IIelp PF3/PF15=Prev Menu PF4/PF16=Main Menu
Figure 4-9. Bar Chart Attribute Selection Menu.
41 4. TUTORIAL
-------
The Bar Chart Attribute Selection Menu allows you to select the pattern and color of the bars
in the bar chart. The values shown in Figure 4-9 (S, RED and S, GREEN) will result in a
solid red bar for the observed concentrations and a solid green bar for the predicted concen-
trations. This illustrates the point that, by changing colors from one bar to the next, you may
want to specify the same pattern for both bars. On the other hand, you may on occasion wish
to change the pattern from one bar to the next while specifying the same color for both bars.
You may also change both the pattern and the color for both bars. The choice is yours to
make.
You will notice that the Bar Chart Attribute Selection Menu (Figure 4-9) presents fields
without presenting lists to accompany them. You may view the list that is available for a
particular field by typing a question mark (?) in the field of interest and then pressing . A window with a list of possible choices will appear. Figure 4-10 illustrates this
window as it will appear in the menu.
The cursor will be on the first item in the list. To make a selection from the list, use the
or cursor movement keys to move the cursor to the item of your choice and press
. The window will disappear and the item you have chosen will appear in the
selection list. Any field may be changed at any time as long as the menu is displayed on
your screen. You may also return to any selection list as many times as you wish by using
the sequence.
The Select Data window only displays the most popular options for each field. For a
complete list of options for each field, refer to the SAS documentation discussed in
Section 5.7 of this manual.
2.4.3
UAMPES: Bai Chart Attribute Selection
Bar
Pattern
Observed Values : £!_
Predicted Values : S_
Enter "?" for a 1 Lst
+Select Data
Command ===>
SELECT ONE
BLUE
CYAN
GREEN
PINK
RED
WHITE
YELLOW
PF'l/PF13=Help
PF3/PFl5=Prev M
Figure 4-10.
Bar Chart Attribute Selection Menu with a Select Data window.
4. TUTORIAL
42
-------
4.4.6 Selecting Titles and Descriptions
After making entries in the Bar Chart Attribute Selection Menu, press to bring up the
Graphic Title/Description Specification Menu shown in Figure 4-11. This menu presents a
table of the text information that will accompany the graphic. Unlike most UAMPES fields,
the fields for title text and axis label text will not be saved. This menu will appear with
default titles and axis labels that accurately describe the graph based on information in the
data sets you specified earlier in the Data File Selection Menu and previous menu fields. The
UAMPES automatically searches the specified data sets and inserts the species, study (defined
by the user when running the Extract module), date, and hours of the data being analyzed.
You may change the wording of these descriptions, but your choices are not saved and will
not reappear in subsequent iterations of this menu.
You may also change the color, font, and size of the lines of text. Colors may be any that
are supported by the device you are using. Fonts may be any of those listed in Chapter 6 of
SAS/GRAPH Software, Version 6 (SAS Institute, Inc., 1990). Be aware that as the size of
text increases, the size of the graph decreases; and as the size of text decreases, the text
becomes less legible. Until you have run the UAMPES once or twice, we recommend that
you use a value of "3" in the title Size fields and a value of "2" in the axis label Size fields.
Similar to most other fields, the Color, Font, and Size fields will be saved and recalled in
subsequent iterations.
Note that the Color, Font, and Size specified for the Horizontal Axis label control the color,
font, and size of the bar labels. You may increase the number of bar labels that can fit on a
single graph by specifying a smaller Horizontal Axis label Size in this menu. The Color
specified for Title #1 will also be used to color the boxes that surround the graph and the
legend, as well as the reference lines and other annotations that appear in the graph. The
Color specified for Title #2 will also be used to color the text that appears in the legend.
r 2.5
UAMPES: Graphic Title/Description Specification
Enter Requested information. Skip any fields not desired. Do not use quote
TEXT COLOR FONT SIZE
Title #1
03 VALUES FOR NEW YORK, 11JUL88 WHITE SIMPLEX 3.00
Title #2
(WITH PERCENT DIFFERENCE ANNOTATED) WHITE SIMPLEX 3.00
Vertical Axis
CONCENTRATION (PPB) WHITE SIMPLEX 2.00
Horizontal Axis
WHITE SIMPLEX 2.00
Enter "?" for a list of valid entries.
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
Figure 4-11. Graphic Title/Description Specification Menu.
43 4. TUTORIAL
-------
4.4.7 Axis Scaling Menu
After you are satisfied with the Graphic Title/Description Specification Menu and press
, the Axis Scaling Menu will appear. This menu allows you to select maximum and
minimum scale values for the vertical axis. If all the graphed values range between 500 and
600, it may be desirable to specify a minimum scale value of 400 in order to take a closer
look at the details of the graph. Also, if you are creating several graphs to be compared with
each other, this menu allows you to ensure that all the graphs use the same scale.
The system checks the data about to be plotted and reports the maximum and minimum Data
Values in the first column of this menu. The second column provides selection fields for you
to specify the maximum and minimum axis values (which should always encompass the full
range of the data values) and the interval to be indicated along the length of the scale line. In
the example shown in Figure 4-12, a Maximum axis value of "200", a Minimum axis value
of "0", and an Interval of "10" have been entered. This interval will result in about 20
labeled tic marks along the scale line. The first time you run the system these selection fields
will be blank. In subsequent runs, these fields will hold the values used on the previous run.
Always check the new data values against the previous axis values to tie sure that the axis
values encompass the full range of data values. If the data values fall outside (either above
or below) the range of axis values, an error message to that effect will be displayed and no
graph will be produced.
r-2.6
UAMPES: BAR CHART AXIS SCALING MENU
Selected
Data Values Axis Values
Maximum 199 200
Minimum 66.26 0
Interval 10
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
Figure 4-12. Bar Chart Axis Scaling Menu.
4. TUTORIAL 44
-------
At this point, you have given the UAMPES all the information necessary to produce a table
of summary statistics and a graph. When you press , a one or two minute pause will
occur while the UAMPES generates the graph. After the graph has been generated, it will be
displayed on your screen. Now, it is time to check the graph to see if it plots the information
in which you are interested, and to check the text describing the graph to see if the default
description is adequate. The table of summary statistics has been placed in your user-
specified output listing file. You may exit the system (UAMPES) and view or print the table
using standard IBM utilities or TSO commands.
If you have faithfully and successfully followed this tutorial, you should see a graph similar
to the one in Figure 4-13 below. The graph displays a pair of bars (one for observed
concentration and one for predicted concentration) for each monitor. The height of each bar
is proportional to the concentration level. The percent difference between the observed and
predicted concentrations is calculated as: [(observed - predicted) / (observed)] x 100, and is
displayed above the bar for the predicted concentration level.
03 VALUES FOR NEW YORK, 11JUL88
(WITH PERCENT DIFFERENCE ANNOTATED)
HOUR=16
36931
361(13
3(119
8AR1
Figure 4-13.
Bar chart (BAR1) from the tutorial.
45
4. TUTORIAL
-------
4.4.8 Termination Options
After you have checked the graph to be sure it is what you had in mind, you must choose
whether or not to save the graph to a graphics catalog. If the graph is just for your informa-
tion or is not exactly what you had in mind, you may be satisfied with simply viewing the
graph on your screen and not saving it. However, if the graph presents information you wish
to share with others, you will probably want to save it to a graphics catalog. Graphs saved to
a graphics catalog can then be displayed on a variety of devices other than the one used to
generate the original graph. Reproducing the graph on a screen or hard-copy device will also
take less time than generating the original graph.
Press to move on to the next menu, the Termination Selection Menu, shown in
Figure 4-14 below. This menu presents five options for the disposition of the graph you have
just viewed. These options will:
(1) return you to the Graphic Selection Menu,
(2) return you to the Data File Selection Menu,
(3) return you to the Graphic Title/Description Menu,
(4) bring up the Graphics Catalog Generation Menu, or
(5) end your current session with the Graphics module option and return to the
UAMPES Main Menu.
For any option, type the corresponding option number in the Selection field and press
. For the purposes of this tutorial, assume that we want to save the graphic. Type
"4" in the Selection field and press .
1-2.7
UAMPES: Termination Selection
Choose the action you would like to take.
1) Change Device or Plot Type
2} Change Input Data Sets or Starting/Ending Conditions
3} Change Titles or Redisplay the Graphic
4) Save the Graphic
5) Exit UAMPES Graphics Module (Without Saving Graphic)
Selection: 4
Figure 4-14. Termination Selection Menu.
4. TUTORIAL 46
-------
4.5 SAVING THE GRAPHIC IN A GRAPHICS CATALOG
The Graphics Catalog Generation Menu shown in Figure 4-15 contains brief instructions for
providing a name for the catalog as well as a name and brief description of the graph that you
have just produced. A single catalog may contain any number of graphics, so you may only
need a single catalog for all the graphics you produce. On the other hand, you may wish to
have more than one catalog for organizational purposes.
The Graphics Catalog DSN should be any valid file name for the computer system that you
are using. On the EPA's IBM system you may wish to use a three-level name beginning with
your user ID/account number (uidacct). Each level must be separated by a period (.) and
should be no more than eight characters in length. Any combination of alphabetical and
numerical characters can be used, but you must start with an alphabetical character (e.g.,
"uidacct.GRAPHICS.CATALOG"). Refer to the Guide to NCC Services (EPA, 1990b) for
further naming conventions.
The graphic Name should be no more than eight alphanumeric characters containing no
blanks, for example, "FIRSTBAR". The graphic Description may be up to 40 characters
long and may contain blanks (e.g., "NEW YORK, 11JUL88, 9 MONITORS, HOUR 16".
r2.7.1
UAMPES: Graphics Catalog Generation
Please enter below the Fully Qualified Data Set Name (DSN)
of the Graphics Catalog to be created or modified.
Do not enclose the Data Set Name in quotes.
Graphics Catalog DSN: uidacct.GRAPHICS.CATALOG
Enter the Name and Description of this graphic. This
information will be stored in the Giaphics Catalog and
will aid in identifying the plot.
Name: FIRSTBAR
Description: NEW YORK, 11JUL.88, 9 MONITORS, HOUR 16
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
Figure 4-15. Graphics Catalog Generation Menu.
47 4. TUTORIAL
-------
After you have provided names and a description in the appropriate fields, press .
The PROC GREPLAY window shown in Figure 4-16 will appear to allow you to verify that
the graphic has been placed in the graphics catalog. Press or to return to the
Termination Selection Menu, shown in Figure 4-14.
Then, you have several choices: produce a different graphic by selecting options 1 or 2; take
another look at the current graphic by selecting option 3; save another copy of the current
graph (possibly to a second catalog) by selecting option 4 again; or end the session by
selecting option 5.
At this point, you should have successfully produced a bar chart and saved it in a graphics
catalog. Refer to Appendix D of this manual for more detailed instructions on using the
"Graphics Catalog" option in the UAMPES Main Menu to view or print graphics stored in a
graphics catalog; this appendix is identical in content to Section 4.3 of the user's manual for
the UAMPPS (EPA, 1990a).
Section 5 (Menu Reference) provides a complete discussion of all UAMPES menus. Section
6 (Files) presents a description of the tables that support the graphics.
PROC GREPLAY
Command ===>
IGOUT: GOUT.PLOTS GOUT: Device: TEK4207
TC: Template: Scroll: PAGE
CC: Cmap:
Sel Name Type Description Created
FIRSTBAR I NEW YORK, 11JUL88, 9 MONITORS, HOUR 16 06/17/92
Figure 4-16. PROC GREPLAY window.
4. TUTORIAL 48
-------
5. MENU REFERENCE
This section describes the UAMPES graphics menus. It begins with an outline of the menu
structure, illustrating how menus are related to each other and the order in which they occur
as you use the UAMPES. The remainder of the section gives detailed descriptions of the
menus. Each description explains a menu's purpose, tells how to use it, and lists error
messages and tells what causes them.
5.1 MENU OUTLINE
The UAMPES menus are numbered in the upper-left corner to give you an idea of where you
are in the system. There are three menu groups that may be chosen from the UAMPES Main
Menu: Extract, Graphics, and Graphics Catalog. All menu numbers associated with Extract
begin with a "1", all menus associated with Graphics begin with a "2", and all menus
associated with the Graphics Catalog begin with a "3".
The following subsections provide information on Graphics menus (i.e., menus stalling with
"2"). As noted earlier in this manual, Appendices A and C present information on the Extract
and Graphics Catalog menus, respectively. These appendices are identical in content to
Sections 4.1 and 4.3 of the user's manual for the UAM postprocessing system (EPA, 1990a)
and are provided herein for easy reference.
The second digit in the menu number indicates the level of the menu. These levels corre-
spond to the levels indicated in Figure 5-1. If there is more than one menu in a given level,
the menus in that level are numbered consecutively with a third digit. For example, using
Figure 5-1, the Bar Chart Attribute Selection Menu will be numbered 2.4.3 (i.e., Graphics
group, level 4, menu 3).
The UAMPES graphics menus allow you to build a variety of graphs using data specified by
you. Figure 5-1 illustrates how the graphics menus are related to each other and the order in
which they appear on your screen.
The Graphic Selection Menu is the starting point for building any type of graph. After
specifying the graphic you want from the Graphic Selection Menu, a menu unique to the type
of graph specified, will come on screen (e.g., Bar Chart Method/Type Selection Menu). This
menu will allow you to further define the type of graph to be produced.
49 5. MENU REFERENCE
-------
,evel 1
Level 2
Level 3
Level 4
2
Level 7
Legend: -
Graphics
Selection
1
2. 1. 1 V 2.1.2V 2,1,3V 2rl,4 V 2.1.5
Spatial TiMe Bar Scatter Qua
Graphics Series Chart Plot P
Method Method Method Method fie
Type Type Type Type T
L.-J L__J L.-I L,J L
22 1
L
Ki
HIP *
rile i
Monitor
<- . . I i t +
LlSl
2.4.IAV 2.4.2V 2.4.3V 2.4.4V 2.4.5
Spatial line Bar Scatter Qua
Graphics Series Chart Plot P
Attribute Attribute Attribute Attribute Att
i
i
2.4..V * I
If 2.5 V
Statistical! i L
f"r,Mr,ne i 4- i nr. J T IP Ti' + la/ I/".
toMposition---i Tonal S litie/ K,
| Level J Description 1
A Tpnol C I"* ?.s V
,4,i.iV LiCVCl 0 «
/ \ Axis
JjS J I Graphic p" Sciling
2.7.1 2.7 V
Graphics a Terwination I
Primary Path = Optional Path = Menu
1 ,,J
itile Bias
ot Plot
thod Method
jpe Type
] ,..,*
ntile Bias
lot Plot
ribute Attribute
i
,1
J-. .__.___.___. N_. ........
1 - Graphic Display
Figure 5-1. The UAMPES menu structure.
5, MENU REFERENCE
50
-------
Next, the Data File Selection Menu and the Print File Selection window, which are common
to all the graphs, will be displayed. These screens let you tell the UAMPES exactly what
data you wish to analyze and where to place the table of summary statistics. If you ask for a
subset of monitors, these screens will be followed by a Monitors Located Within Domain
Menu which is also common to all the graphs. If you ask to use all the monitors or a
previous subset of monitors, you will skip the Monitors Located Within Domain Menu and
proceed directly to the Attribute Selection Menu(s) unique to the type of graph selected. Then
you can select specific attributes for the graph you are creating.
The Graphic Title/Description Specification Menu follows the specific Attribute Selection
Menu; this menu is common to all the graphs. It lets you specify the text that describes the
graph. In the case of the spatial and the bias plot graphics, when this menu is completed, the
table of summary statistics and the graph are completely defined and will be generated. The
table of summary statistics is placed in the print file and the graph is displayed on your
screen. For the other four graphics (time series, bar chart, scatter plot, and quantile plot), the
Axis Scaling Menu will be presented before the graph is displayed. The Axis Scaling Menu
allows you to specify the upper and lower limits and the interval for the scaled axis or axes.
After the table of summary statistics is placed in the print file and the graph is displayed, the
Termination Selection Menu presents several options for what to do next. If you choose to
save the graphic, the Graphics Catalog Generation Menu will prompt you for the name of the
catalog, and a name and description for the graph to be saved. After saving the graphic, the
Termination Selection Menu will reappear. At this point, you may generate another graphic
or end the session. After ending the session, you may use standard IBM or TSO commands
to view or print the print file.
The remainder of this section describes each of the UAMPES graphics menus in detail.
51 5. MENU REFERENCE
-------
5.2 UAMPES MAIN MENU
The UAMPES Main Menu (Figure 5-2) presents five possible options: the Extract module,
the Graphics module, the Graphics Catalog module, Help, and Exit. The Extract and
Graphics Catalog options are identical to the Extract and Graphics Catalog options in the
UAMPPS. Briefly, the Extract option allows users to extract data for a single species at a
single level from a much larger data set generated by running the Urban Airshed Model. The
Graphics Catalog option allows users to view or produce hard-copy output of graphs that have
been saved in a graphics catalog. Please refer to Appendices A and D for details of the
operation of these options. The Graphics option is the focus of this section. The Help option
provides general information about the system and the other options available in the UAMPES
Main Menu. The Exit option will end the current session and return you to the TSO READY
prompt.
When you first start the system, your cursor will appear on the Extract option. Press the
key to move the cursor from one option to another. To select a particular option,
move the cursor to the desired option and press . To view and work with the menus
discussed in this section, move the cursor to the Graphics option and press .
URBAN AIRSHED MODEL PERFORMANCE EUALUATION SYSTEM"
MAIN MENU
Place cursor on your selection and press the enter key.
Figure 5-2. UAMPES Main Menu.
5. MENU REFERENCE
52
-------
5.3 GRAPHIC SELECTION MENU
The UAMPES Graphic Selection Menu (2.0), shown in Figure 5-3, lets you select a graphic.
In addition, the menu asks for the type of equipment you are using so that the system knows
what language to "speak" when it sends the graph to your terminal.
r2.0
URBAN AIRSHED MODEL PERFORMANCE EVALUATION SYSTEM (UAMPES)
Graphic Selection
1) Spatial Graphic
2) Time-Series Line Graph
3) Bar Chart
4} Scatter Plot
5} Quantile Plot
6) Bias Plot
Graphic Selection: _ Device:
Enter "?" on Device field for a list of supported devices.
PFl/PF13=Help PF3/PF15=Exit PF4/PF16=Main Menu
Figure 5-3. Graphic Selection Menu.
Instructions
Choose a graphic from the list and enter the number of that graphic in the Graphic Selection
field. Samples of the available graphics can be found in Section 2 of this manual.
Next, enter the name of the terminal you are using or emulating in the Device field. Be sure
to consult the user's manual supplied by the terminal manufacturer if you are not certain what
type of graphics device you are using or emulating. For further help, call the EPA User
Support Department (see Section 1).
Note that the UAMPES only checks the Device field for a supported device. There is no way
for the UAMPES to know if the device you have entered accurately describes the one you are
using so there will NOT be an error message if the entry is incorrect. There will ONLY be
an error message if the entry does not match one of the supported devices. Failure to enter
the proper device name on this screen may result in unrecoverable errors when you attempt
to display a graph.
53 5. MENU REFERENCE
-------
If you are using a Tektronix 4207, 4211, or 4224 connected via a coaxial adapter, enter
"TCX4207". If you are using one of the above named devices but NOT with a coaxial
connection, enter "TEK4207". Enter "IBM3179" or "GDDMPCG" if you are using or
emulating an IBM 3179G terminal; "IBM3279" should be entered if you are using or
emulating an IBM 3279G terminal. Enter "TEK4010" if you are using a version of Kermit
that supports graphics. After you have made valid entries in the Graphic Selection and
Device fields, press to move on to the next menu.
Error Messages
A value is required at the cursor location.
Because both fields in this menu are required, leaving either of them blank will result in the
above message appearing in the Error Message window. Press to remove the error
window from your screen. The cursor will be placed on the appropriate field to allow you to
enter a valid value. Use the guidelines described above to supply the correct information.
You must enter a value between to continue.
The above message will be displayed if any value other than "1" through "6" is entered in the
Graphic Selection field. The cursor will be placed on the Graphic Selection field to allow
you to enter a number within the range of 1 through 6.
Finally, if you specify a value in the Device field that is not in the list of supported terminal
devices, a Device List Window will open and you may choose a device from the list.
5. MENU REFERENCE 54
-------
5.4 METHOD/TYPE SELECTION MENUS
There is a Method/Type Selection Menu for each type of graph (UAMPES menus 2.1.1
through 2.1.6) that can be specified in the UAMPES Graphic Selection Menu. The menu for
the spatial graphics has unique options because it draws a map of the geographical area in
question. Menus for the other graphics are quite similar to each other. Each of the other five
menus has three similar lists to choose from. Accordingly, the Spatial Graphics Annotation
Selection Menu will be discussed separately followed by a collective discussion of the other
Method/Type Selection Menus.
5.4.1 Spatial Graphics Annotation Selection Menu
The Spatial Graphics Annotation Selection Menu (2.1.1), shown in Figure 5-4 below, lets you
select the type of spatial graphic that you would like to create. You can display either color
or monochrome versions of a shaded-tile or gridded-value map, using one of the four
Analysis types.
12.1.1-
UAMPES: Spatial Graphics Annotation Selection
Map Type:
Pattern Type:
Grid Annotation:
City Annotation:
Monitor Selection':
Analysis:
PFl/PF13=Help
1) Shaded Tile
1) Color
1) None
2) Gridded Value
2) Monochrome
2) Col/Row
3) Lat/Lon
1} None 2) All 3} Subset Cities 4} Prev Subset
1) All Monitors 2) Subset Monitors 3} Prev Subset
1) Average Concentration
2) Maximum Concentration
3) Hour of Maximum Concentration
4} Concentration at a Given Hour
PF3/PF15-Prev Menu
PF4/PF16=Main Menu
Figure 5-4. Spatial Graphics Annotation Selection Menu.
55
5. MENU REFERENCE
-------
Instructions
Choose either a shaded-tile or gridded-value map by specifying "1" or "2" in the Map Type
field. After selecting the type of graphic that you would like to display, you may then choose
to display either a color or monochrome plot. Inserting "1" in the Pattern Type field will
yield a color plot; a value of "2" will generate a monochrome plot.
Next, you must decide the type of grid annotation that you want displayed on the chart.
Entering "1" in the Grid Annotation field will display no annotation. Entering "2" will
annotate row and column numbers. Entering "3" will display latitude/longitude annotation.
The City Annotation field lets you determine what cities, if any, will be marked on the map.
A value of "1" displays no cities. A value of "2" displays all major cities within the domain.
However, you may find that displaying all cities within the domain produces a cluttered map.
If you enter "3", you will be able to select the cities that you would like displayed from the
complete list. After you have decided on a subset of cities to display, you may repeat that
subset of cities on subsequent runs by entering "4". Note that you MUST enter "3" before
entering "4". Also, the subset is not retained at the end of a session: Specifically, if you
change domains within a session, you must enter "3" to produce a valid subset for the new
domain. For example, entering "4" to obtain a previous subset for New York would not
overlay on a domain for Atlanta.
The Monitor Selection field is similar to the City Annotation field. It lets you determine
which monitors to display on top of the grid cells. A value of " 1" will display all the
monitors in the selected domain. Again, displaying all the monitors may produce a cluttered
map. If you enter "2", you will be able to select the monitors to display from the complete
list. After selecting a subset of monitors, you may repeat that subset of monitors on
subsequent runs by selecting "3". The same caveats that apply to selections "3" and "4" of
the City Annotation field also apply to selections "2" and "3" of the Monitor Selection field
(i.e., a value of "2" must be run before entering "3", and if you change domains, "3" is not
available).
The Analysis field (chosen next) that is applied to the grid cells will also be applied to any
monitors that are displayed. If you choose to display several monitors and select "1" for the
average concentration analysis, then the average concentration of each monitor over the
selected range of hours will be displayed at that monitor's location. If you have selected a
shaded tile map (Map Type "1"), then the monitor location will be shaded accordingly. If
you have selected a gridded value map (Map Type "2"), then the monitor value (average,
maximum, etc.) will be displayed at the monitor location.
By filling out the Analysis field, you determine the type of statistic that is displayed in the
graphic. The types of statistics and their corresponding options are explained below.
5. MENU REFERENCE 56
-------
Option IAverage Concentration
Computes the average concentration for each cell in the domain and the
average concentration for each monitor. This average is based on the number
of hours specified in the Data File Selection Menu. The range of hours
specified is considered to be the period of interest.
Option 2--Maximum Concentration
Computes the maximum concentration for each cell in the domain and the maximum
concentration for each monitor over the specified time period and displays the result in
the graphic.
Option 3-Hour of Maximum Concentration
Determines the hour at which the maximum predicted concentration for each
grid cell and maximum observed concentration for each monitor in the domain
occurs over the period of interest, and displays the result in the graphic.
Option 4--Concentration at a Given Hour
Determines the concentration at the hour specified in the Start Hour field of
the Data File Selection Menu and displays the result in the graphic.
After you have specified all information, press to process your selections.
Error Messages
Enter a number between -or- press PF1/PP13 for help.
The above message appears if the number in any field is not in the allowable range for that
field. Type over the incorrect value with a valid value.
Selection 4 only valid after selection 3 has been run.
The above message results from typing a "4" in the City Annotation field, when there is no
previously selected subset of cities. Always choose option 3 before choosing option 4.
During a session, you must create a subset before you use it.
57 5. MENU REFERENCE
-------
5.4.2 Other Method/Type Selection Menus
For the other five graphics (other than spatial graphics), the selection menus (2.1.2 through
2.1.6) display different selection lists for each of three fields: Type, Aggregation Method,
and Monitor Selection. The Time-Series Method/Type Selection Menu (2.1.2), shown in
Figure 5-5 below, is representative of these menus. The choices for graphic Type are listed
first, followed by a list of Aggregation Method choices. The features and algorithms
associated with each type and method are discussed in Section 2 and Appendix C. Then, the
user chooses the Monitor Selection (all or a subset of monitors).
-2.1.2
UAMPES: TIME-SERIES METHOD/TYPE SELECTION
Type: _ 1. Observed and Predicted (Select Aggregate)
2. Maximum Observed and Predicted (Select Aggregate)
3. Domain Maximum Observed and Predicted
Aggregation Method: _ 1. Samecell
2. 9 Cell Closest Value
3. Distance Weighted 9 Cell Average
4. 4 Cell Closest Value
5. Distance Weighted 4 Cell Average
6. Bilinear 4 Cell
Monitor Selection: _ 1. Use All Monitors
2. Subset Monitors
3. Use Previous Subset of Monitors
PF/l/PF13=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
Figure 5-5. Time-Series Method/Type Selection Menu.
Instructions
Choose a graphic type from the list and enter its number in the Type field. Do the same for
the aggregation method, entering the appropriate number in the Aggregation Method field.
Choosing all or only a subset of monitors will have different effects depending on the type of
graph that you desire. Both the Aggregation Method selected and the Monitor Selection
will have a direct bearing on the table of summary statistics. The Type selected will have no
effect on the table.
The first Type of time-series graphics, Observed and Predicted (Select Aggregate), will
generate a separate graph for each monitor. You may wish to select only a single monitor
and wait to see the corresponding graph before electing to generate multiple graphs which
may or may not contain useful information.
5. MENU REFERENCE 58
-------
Using a subset of monitors will primarily affect the appearance of the bar charts and scatter
plots by plotting fewer bars and dots respectively. See the tutorial in Section 4 for a
thorough discussion of other considerations affecting bar charts.
In addition, using a subset of monitors will affect the calculations of rank order in the
quantile plots and fractional bias in the bias plots. Refer to Section 2 and Appendix C of this
manual for a discussion of these calculations.
After you have made valid entries in all three fields, press to move on to the next
menu.
Error Messages
Error messages for these screens are the same as those for the UAMPES Graphic Selection
Menu.
59 5. MENU REFERENCE
-------
5.5 DATA FILE SELECTION MENU AND PRINT FILE SELECTION WINDOW
The Data File Selection Menu lets you specify exactly which data sets you want the UAM-
PES to compare. The Print File Selection window allows you to direct the text output that
accompanies each graph produced by the UAMPES. The Print File Selection window will
appear only once each time you run the UAMPES, and will appear on top of the Data File
Selection Menu the first time that menu appears in a session.
5.5.1 Data File Selection Menu
The Data File Selection Menu (2.2), shown in Figure 5-6, contains five selection fields that
must be completed before moving on to the next menu. There are two long fields for
entering data set names. These long fields are followed by three shorter fields for entering
the date, and start and end hours.
2.2
UAMPES: Data File Selection
Please enter the Fully Qualified Data Set Names for the Extracted
UAM (predicted) data set and the AIRS AMP350 (observed) Workfile
to be used in this graphic.
Do not enclose the Data Set Name in quotes.
Extracted Data Set:
AIRS AMP350 Workfile:
Period of Interest : Date Start Hour
End Hour
(DDMMMYY) (1-24;
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
Figure 5-6. Data File Selection Menu.
Instructions
The first long field calls for the name of an extracted data set created by the Extract module.
Enter the fully qualified data set name of the extracted data set in the space provided. The
second long field calls for the name of an AIRS AMP350 Workfile. Enter the fully qualified
data set name of the workfile data set in the space provided. Do not enclose the data set
names in quotation marks.
5. MENU REFERENCE 60
-------
The three short selection fields define the period of interest and consist of the date and
starting and ending hours. The period of interest must be within the time frame contained in
the two data sets. The Date should contain seven characters of the form ddmmmyy where dd
is the day, mmm is the month, and yy is the year. For example, if the extracted and workfile
data sets contain data for hours 1 to 24 on August 1, 1991, then you must enter "01AUG91"
in the Date field and may choose any range of hours between 1 and 24 for the Start and End
Hour fields. The only restriction is that the starting hour must be earlier than the ending
hour.
Error Messages
Data set not found.
The UAMPES will search for data sets with the names that you have specified. If they are
not found, the message above will appear. Press to remove the Error Message
window from the screen. The data set that was not found will be highlighted. If neither data
set was found, both will be highlighted. You should first check the entries you have made in
this menu for accuracy. If you do not find any errors, end the current session and review the
list of data sets. To end the current session, press to return to the UAMPES
Main Menu and press to exit the system.
Ending time precedes starting time.
This message will appear if the ending time occurs earlier than the starting time. Change
either the starting or ending time so that the analysis starts before it ends. The program will
not run if you start the analysis at 12 and end it at 10.
You, the user, are responsible for determining that both of the data sets consider the same
species of pollutant. Obviously, it makes no sense to compare predicted levels of ozone with
observed levels of carbon monoxide. Check the data sets you intend to analyze before you
start the system to be sure that they contain values for the same species of pollutant. This is
one instance in which you can ask for nonsense and the system will let you have it.
61 5. MENU REFERENCE
-------
5.5.2 Print File Selection Window
For each graph produced by the UAMPES, there will be a corresponding text file produced
that records the exact values or data points plotted in each graph and contains the summary
statistics table (Figure 2-1, Section 2.2). The summary statistics table is generated every time
the user selects a graph for which an aggregation method is specified. The text file will
duplicate the titles and label of the graph it accompanies. The Print File Selection window,
shown in Figure 5-7, allows you to tell the system what to name this text file. Choose any
file name that is a valid name for your computer system. If your system requires file
allocation, UAMPES will allocate the file if it has not been previously allocated. The Print
File Selection window will appear on top of the Data File Selection Menu. After each graph
or group of graphs you generate, various options allow you to return to the Data File
Selection Menu or to an earlier menu to generate multiple graphs or groups of graphs within
the same session. Thus, it is possible to be presented with the Data File Selection Menu
several times in a single session. The Print File Selection window will only appear the first
time through the Data File Selection Menu. Text files will be appended to each other until
you end the session by exiting the system.
After you run the UAMPES the first time, the Report File Name field will be saved and
recalled just like most of the other fields in the UAMPES. If you use the same file name in a
subsequent session, the file from the previous session will be over written. This is intended
as a convenience to the user as a method of conserving computer resources. If you do not
want the previous file to be over written, either rename the existing file before starting the
UAMPES, or specify a new name in the Print File Selection window.
1-2.2
UAMPES: Data File Selection
+-Press ENTER to Continue
Print File Selection
Please enter the Fully Qualified Data Set Name for the text file
that will receive all report output for this session.
Do not enclose the Data Set Name in quotes.
Report File Name:
Be sure to record the file name for later use.
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
Figure 5-7. Data File Selection Menu with Print File Selection window.
5, MENU REFERENCE 62
-------
5.6 MONITOR SELECTION MENU
The Monitors Located Within Domain Menu (2.3), shown in Figure 5-8, will be presented
only if you select the second Monitor Selection option (subset monitors) in the Method/Type
Selection Menu (Section 5.4).
r 2.3
Monitors Located Within Domain
Place an S in the selection column for all monitors desired in this run.
Press PF20 to scroll down the list, PF19 to scroll up.
After all selections have been made, press PF15 to save selections.
Select Bar Label Monitoi ID State/County/Site Address
_ 13007 090013007 CT/FAIkFJELD/USCG LIGHTHOUSE, PROSPECT
31003 090031003 CT/HARTFORD/30 REMINGTON ROAD
_ 70007 090070007 CT/M1DDLESEX/CONN. VALLEY HOSP. , SHEW
91123 090091123 CT/NEW HAVEN/715 STATE STREET
_ 31001 090131001 CT/TOLLAND/ROUTE 190, SHENIPSIT STATE
70006 340170006 NJ/HUDSON/VETERANS PARK ON NEWARK BAY
10005 340210005 NJ/MERCER/RIDER COLLEGE;LAWRENCE TOWNS
_ 30006 340230006 NJ/MIDDLESEX/RYDERS LANE & LOG CABIN R
10004 360810004 NY/QUEENS/QUEENS COLLEGE, 65-20 KISSEN
30002 361030002 NY/SUFFOLK/EAST FARMINGDALE WATER DIST
92004 361192004 NY/WESTCHESTER/WHITE PLAINS PUMP STATI
Figure 5-8.
Monitor* Located Within Domain Menu.
Before presenting the list that appears in this menu, the UAMPES compares the two data sets
entered in the Data File Selection Menu, and matches the monitors and their locations in the
AIRS AMP350 workfile with the geographical area described by the domain in the extracted
data set. Only the monitors that are located within the domain of the extracted data set are
included in the UAMPES analysis. The monitors that do fit this criteria are then listed in the
Monitors Located Within Domain Menu.
The Monitors Located Within Domain Menu lets you preview the group of monitors that will
be used in the model performance analysis. You may then select all of the monitors, a single
monitor, or any subset of monitors you wish. The monitor list consists of four columns of
information. The first column is the Select field. The second column, the Bar Label field,
is an abbreviated identification. By default, the Bar Label is composed of the last five
characters of the monitor ID. The third column is the Monitor ID field. The fourth column
describes the location of the monitor in terms of State, County, and Site Address. The State
63
5. MENU REFERENCE
-------
is listed in the standard two-letter abbreviation followed by the county and then as many
characters of the site address (as listed in the AMP350 workfile) as will fit on the screen.
Instructions
Only 11 monitors can be listed on the screen at one time. If there are more than 11 monitors
to choose from, you will need to use the and keys to scroll the list.
Use to move the cursor to the Select field of each monitor that you wish to include in
this graph. Type an "S" to indicate that this monitor is to be included. If you type an "S"
and then decide you don't want this monitor included in this graph, return the cursor to the
Select field using the key and press the to erase the "S". Selections
remain while you scroll through the list.
The Bar Label column is used only in composing bar charts. Refer to Section 4.4.5 for a
discussion of how to use this column if you are generating a bar chart. After you have
selected all the monitors that you want for the current analysis (and you are satisfied with the
Bar Label column if generating a bar chart), press the key to save the selections
you have made and move on to the next menu. The key takes the place of
in this menu because the selections to be saved may include selections that have
been scrolled off of the screen.
Error Messages
There are no error messages associated with this menu. The system will read any character
that appears in the Select field as though it were an "S".
5. MENU REFERENCE 64
-------
5.7 ATTRIBUTE SELECTION MENUS
Each of the six graphics groups has a corresponding Attribute Selection Menu (UAMPES
menus 2.4.1, through 2.4.6) that determines certain appearance attributes for that particular
graphic. These graphic-specific Attribute Selection Menus are discussed separately, in the
following subsections.
You will notice that these menus present selection fields without presenting selection lists to
accompany them. You may view the list that is available for a particular selection field by
typing a question mark (?) in the field of interest and then pressing . A window with
a list of possible choices will appear (e.g., see Figure 4-10 in Section 4.4.5).
When the window appears, the cursor will be on the first item in the list. To make a
selection from the list, use the or cursor movement keys to move the cursor to the
item of your choice and press . The window will disappear and the item that you
have chosen will appear in the selection field. Any field may be changed at any time as long
as the menu is displayed on your screen. You may also return to any selection list as many
times as you wish using the sequence.
The selection list window only displays the most popular options. For a complete list of
options for each selection field, you will want to refer to the following figures and tables in
the SAS/GRAPH Software, Version 6 (SAS Institute, Inc., 1990).
Fonts, see Table 6.3 Patterns, see Figure 15.4
Symbols, see Figure 6.3 Line types, see Table 16.5
Colors, see Table 7.2.
5.7.1 Spatial Graphics Pattern Selection and Statistical Composition of Data
The Spatial Graphics Pattern Selection Menu (2.4.1) shown in Figure 5-9 lets you select
cutoff values for the class breaks and a corresponding color or monochrome fill pattern to be
displayed on the graphic. The Statistical Composition of Data Menu shown in Figure 5-10 is
the second page of the Spatial Graphics Pattern Selection Menu, and is invoked by pressing
after the Spatial Graphics Pattern Selection Menu is displayed. The purpose of
the Statistical Composition of Data Menu is to provide you with the information necessary to
enter the cutoff values on the Spatial Graphics Pattern Selection Menu. Based on the graphic
chosen on the Graphic Selection Menu (Figure 5-3), you will be presented with the values of
the data to be plotted, broken down into several percentiles. Thus, you can determine the
approximate percent of the data that will be contained in the cutoff ranges you select.
65 5. MENU REFERENCE
-------
^ . <± . Irt
Spatial Graphics Pattern Selection
Press PF20 to see the Statistical Composition of the Data.
Select the appropriate class breaks and corresponding colors or patterns.
The available colors and patterns for your device are specified below.
Default Assigned Assigned
Class Cutoff Value Cutoff Value Color/Pattern
1 < 82 < 80 BLUE
2 <
3 <
4 <
5 <
6 >, -
96 < 95 CYAN
110 < 110 GREEN
124 < 125 YELLOW
138 < 140 PINK
138 >,= 140 RED
Available Colors (for Color graphics)
BLUE CYAN GREEN YELLOW PINK RED WHITE
Available Patterns (for Monochrome graphics)
E M1N45 M3N45 M4N45 M5N45 S
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
Figure 5-9. Spatial Graphics Pattern Selection Menu.
4 . t . irs
Statistical Composition of Data
Press PF19 to return to the Pattern Selection
Percentile
5
10
15
20
25
30
35
40
45
50
Minimum Value
Number of Records
PF1/
PF13=Help
Value
81.4
89 .9
94.1
97 .1
99.1
105.1
110.6
115.5
119.4
122.2
68.4
667
PF3/PF15
Percentile
55
60
65
70
75
80
85
90
95
Maximum Value :
=Prev Menu
Menu .
Value
126
129
132
134
138
140
143
145
148
154
3
6
3
8
4
7
5
5
8
5
PF4/PF16=Main Menu
Figure 5-10. Statistical Composition of Data Menu.
S. MENU REFERENCE
66
-------
Instructions
The Default Cutoff Value column offers a suggestion of six cutoff values derived from the
data range detailed on the Statistical Composition of Data Menu. You may enter these values
in the Assigned Cutoff Value field of the Spatial Graphics Selection Menu (Figure 5-9), or
enter your own values. All six fields must have a value, and the values must be in ascending
magnitude from top to bottom. Note that the logical operator preceding the Assigned Cutoff
Value field will be used to subset the data for this class. For example, as shown in
Figure 5-9, the following class breaks are defined:
Class 1 -- Data values less than 80 will be colored blue.
Class 2 - Data values greater than or equal to 80 but less than 95 will be
colored cyan.
Class 3 Data values greater than or equal to 95 but less than 110 will be
colored green.
Class 4 - Data values greater than or equal to 110 but less than 125 will be
colored yellow.
Class 5 - Data values greater than or equal to 125 but less than 140 will be
colored pink.
Class 6 Data values greater than or equal to 140 will be colored red.
You must also enter a color or monochrome fill pattern in the corresponding Assigned Color/
Pattern field. We recommend that you choose from the colors/patterns listed at the bottom
of this screen, because these colors are supported by the device you specified on the Graphic
Selection Menu. Also, the available monochrome patterns (listed) have yielded good results,
therefore they should be used if pattern fills are needed.
Note that the values for Assigned Cutoff Value and Assigned Color/Pattern fields are
retained from the previous run. If you change species, levels, or analysis types, the Assigned
Cutoff Value may not adequately represent the range of values present. Always refer to the
Default Cutoff Value column or the Statistical Composition of Data Menu before selecting
cutoff values.
Error Messages
Value required at cuisor location.
If the Class 1 Cutoff Value field or any Color/Pattern field is left blank, then the above
message will appear. Enter a value at the appropriate location.
Cutoff value <3> must be greater than cutoff value <2>.
Each succeeding Cutoff Value field must be greater than the preceding one. If this message
appears, change the appropriate fields to meet this condition.
67 5. MENU REFERENCE
-------
5.7.2 Cities Within Domain
If City Annotation field selection 3 is chosen on the Spatial Graphics Annotation Selection
Menu (Figure 5-4), then the Cities Located Within Domain Menu (2.4.1.1) shown in
Figure 5-11 will be displayed. It allows you to subset the cities within the domain covered
by the extracted data set.
r2.4.1.1
Cities Located Within Domain
Place an S in the selection column for all cities to be displayed on map.
Press PF20 to scroll down the list, PF19 to scroll up.
After all selections have been made, press PF15 to save selections.
Selection City
*BRIDGEPORT
*BRISTOL
*DANBURY
_ *FAIRFIELD
*GREENWICH
_ *HAMDEN
*HARTFORD
*MANCHESTER
_ *HERIDEN
*MILFORD
*NEW BRITAIN
Figure 5-11. Cities Located Within Domain Menu.
Instructions
All cities included in the SAS cities data set that fall within the current domain will be
present in the list of cities. The list may be several screens long. To scroll down the list,
press the key; to scroll up, press the key. Type an "S" in the
Selection column next to each city that you want in the City subset. After you have
completed the selection process, press to save the subset and move on to the
next menu.
Error Messages
There are no error messages for this menu.
5. MENU REFERENCE 68
-------
5.7.3 Time-Series Attribute Selection
The time-series graph displays symbols that represent the data points being plotted. Each
group of symbols is connected by a line that helps show relationships between the two data
sets. The Time-Series Attribute Selection Menu (2.4.2) shown in Figure 5-12 allows you to
specify line type, line color, symbol type, and symbol size. The most common line types are
solid, dashed, and dotted. Line color also determines symbol color and may be any color
supported by the device you are using. The most common symbol types are squares,
triangles, and stars.
-2.4.2
UAMPES: Time-Series Attribute Selection
Line Line Symbol Symbol
Type Color Type Size
Observed Values:
Predicted Values:
Enter "?" foi a list of valid entries.
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
Figure 5-12. Time-Series Attribute Selection Menu.
Instructions
Enter a "?" in the Line Type, Line Color, and Symbol Type fields to see a list of valid
entries for each field. Within a selection list, to the entry you wish to select and press
. Your selection will appeal' in the selection field. The size of the symbol may be
adjusted to improve clarity or detail. Enter a number between 0 and 10 in the Symbol Size
field. Symbol Size is a three-character field, because you may want to enter decimal values
(e.g., ".75", "2.2").
Error Messages
Please enter a size between .
69 5. MENU REFERENCE
-------
The error message will be displayed if any value other than 0 through 10 is entered in the
Symbol Size field. The cursor will be placed on the Symbol Size field to allow you to enter
a number within the range of 0 through 10.
There are no error messages for the Line Type, Line Color, or Symbol Type fields. If the
entry is not valid, the system will select a default entry and plot the graph accordingly. Note
that system defaults may give unexpected results.
5. MENU REFERENCE 70
-------
5.7.4 Bar Chart Attribute Selection
The Bar Chart Attribute Selection Menu (2.4.3) shown in Figure 5-13 lets you select the
pattern and color of the bars in the bar chart.
r2.4.3
UAMPES: Bar Chart Attribute Selection
Bar Bar-
Pattern Color
Observed Values: S RED
Predicted Values: S GREEN
Enter "?" for a list of valid entries.
PFl/PF13=Help PF3/PFlb=Prev Menu PF4/PF16=Main Menu
Figure 5-13. Bar Chart Attribute Selection Menu.
Instructions
You should enter a "?" in the Bar Pattern and Bar Color fields to see a list of valid entries
for each field. As shown in the example in Figure 5-13, the values "S", "RED" and "S",
"GREEN" will result in a* solid red bar for the observed concentrations and a solid green bar
for the predicted concentrations. Thus, by changing colors from one bar to the next, you may
want to specify the same pattern for both bars. On the other hand, you may on occasion wish
to change the pattern from one bar to the next, but specify the same color for both bars. You
may also change both the pattern and the color for both bars.
Error Messages
There are no error messages for this menu. If the entry is not valid, the system will select a
default entry and plot the graph accordingly. Note that system defaults may give unexpected
results.
71 5. MENU REFERENCE
-------
5.7.5 Scatter Plot Attribute Selection
The Scatter Plot Attribute Selection Menu (2.4.4), shown in Figure 5-14, lets you select
appearance attributes for the symbol that is plotted and adjust the location of the reference
lines in the scatter plot.
r2.4.4
UAMPES: Scatter Plot Attribute Selection
Symbol Symbol Symbol
Color Type Size
Observed vs Predicted:
Enter "?" for a list of valid entries.
Factor Values for Reference Lines
Factor of : 1.30 = 30% Difference
Factor of : 2.00 = 100% Difference
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
Figure 5-14. Scatter Plot Attribute Selection Menu.
Instructions
Select the desired color, type, and size of the symbol. Enter a "?" in the Symbol Color,
Symbol Type, and Symbol Size fields to see a list of valid entries for each field. Within a
selection list, to the entry that you wish to select and press . Your selection
will appear in the selection field. Enter a number between 0 and 10 in the Symbol Size field.
Symbol Size is a three-character field, because you may want to enter decimal values (e.g.,
".75", "2.2").
Reference lines are located by default at 0%, indicating no difference between observed and
predicted values; plus or minus 30%, indicating that observed values are 30% larger or
smaller than predicted values; and plus or minus 100%, indicating that observed values are
100% larger or smaller than predicted values. You may change the two plus or minus
reference lines to any factor value >1 and <51. The 0% reference line is always plotted. Be
sure to enter factor values rather than percentages. The text in the menu reminds you that a
30% difference is the same as 1.3 times larger, or a factor of 1.3 larger; and a 100%
difference is twice as large, or a factor of 2.0 larger. Use these examples to guide you and
5, MENU REFERENCE 72
-------
remember that entering the digits "30" in the selection field will show the location of values
that are 30 times larger rather than 30 percent larger.
Error Messages
Please enter a size between .
The above message will be displayed if any value other than 0 through 10 is entered in the
Symbol Size field. The cursor will be placed on the Symbol Size field to allow you to enter
a number within the range of 0 through 10.
Invalid Factor Value. Please reenter.
The above message will be displayed if you enter a Factor Value less than or equal to 1 or
greater than 51.
There are no error messages for the Symbol Color or Symbol Type fields. If the entry is not
valid, the system will select a default entry and plot the graph accordingly. Note that system
defaults may give unexpected results.
73
5. MENU REFERENCE
-------
5.7.6 Quantile Plot Attribute Selection
The Quantile Plot Attribute Selection Menu (2.4.5), shown in Figure 5-15, lets you select a
color for the symbols and change the location of the reference lines in the quantile plot. This
plot uses empty circles for data points and filled circles for annotated percentiles. Because
the design of this plot is built around empty and filled circles, you are not given an option to
change the Symbol Type.
r2.4.5-
UAMPES: Quantile Plot Attribute Selection
Open Filled
Symbol Symbol Symbol
Color Color Size
Observed vs Predicted:
Enter "?" for a list of valid entries.
Factor Values for Reference Lines
Factor of : 1.30 = 30% Difference _
Factor of : 2.00 = 100% Difference
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
Figure 5-15. Quantile Plot Attribute Selection Menu.
Instructions
Select the desired color and size of the symbol. Enter a "?" in the Open Symbol Color and
Filled Symbol Color fields to see a list of valid entries for each field. Enter a number
between 0 and 10 in the Symbol Size field. These fields operate the same as the fields in the
Scatter Plot Attribute Selection Menu. Please refer to the instructions for Scatter Plot
Attribute Selection Menu (in Section 5.7.5) for details on adjusting reference lines.
Error Messages
The error messages for this menu are the same as those for the Scatter Plot Attribute
Selection Menu (see Section 5.7.5).
5. MENU REFERENCE 74
-------
5.7.7 Bias Plot Attribute Selection
The Bias Plot Attribute Selection Menu (2.4.6), shown in Figure 5-16, lets you select the
appearance attributes for the symbol that is plotted and adjust the location of the reference
lines in the bias plot.
r2.4.6
UAMPES: Bias Plot Attribute Selection
Symbol Symbol
Type Color
Observed vs Predicted:
Enter "?" for a list of valid entries.
Factor Values for Reference Boxes
Factor of : 1.30 = 30% Difference Small Box
2.00 = 100% Difference Large Box
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
Figure 5-16. Bias Plot Attribute Selection Menu.
Instructions
You should enter a "?" in the Symbol Type and Symbol Color fields to see a list of valid
entries for each field. Select the desired type and color of the symbol.
Factor Values are similar to those in the scatter plot and quantile plot except that the values
are applied to reference boxes rather than reference lines. Please refer to the discussion for
the Scatter Plot Attribute Selection Menu (see Section 5.7.5) for details on adjusting reference
lines.
Error Messages
Invalid Factor Value. Please reenter.
The above message will be displayed if you enter a Factor Value less than or equal to 1 or
greater than 51.
There are no error messages for the Symbol Type or Symbol Color fields. If the entry is not
valid, the system will select a default entry and plot the graph accordingly. Note that system
defaults may give unexpected results.
75 5. MENU REFERENCE
-------
5.8 GRAPHIC TITLE/DESCRIPTION SPECIFICATION MENU
The Graphic Title/Description Specification Menu (2.5), shown in Figure 5-17, presents a
table of the text information that will accompany the graphic. This menu will appear with
default titles and axis labels that accurately describe the graph based on information in the
data sets you specified earlier in the Data File Selection Menu and in previous menu entries.
2.5
UAMPES: Graphic Title/Description Specification
Enter Requested information. Skip any fields not desired. Do not use quote
TEXT COLOR FONT SIZE
Title #1
Title #2
Vertical Axis
Horizontal Axi:
Enter "?" for a list of valid entries.
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PFl6=Main Menu
Figure 5-17. Graphic Title/Description Specification Menu.
Instructions
The UAMPES automatically searches the specified data sets and inserts the species, study
(defined by the user when running the Extract module), date, and hours of the data being
analyzed. You may change the wording of these descriptions, but the changes will not be
saved. Each time you run the system, default values for the Text fields will be derived from
the data sets specified in the Data File Selection Menu. This feature allows you to verify the
files specified in that menu.
The values that you enter in the Color, Font, and Size fields will be saved and recalled just
like most other fields in the UAMPES. Colors may be any that are supported by the device
you are using. Fonts may include any listed in Chapter 6 of SAS/GRAPH Software,
Version 6 (SAS Institute, Inc., 1990). Enter a "?" in the Color and Font fields to see a list
of valid entries for each field. Within a selection list, to the entry that you wish to
select and press . Your selection will appear in the selection field. Enter a number
between 0 and 6 in the Size field. Size is a four-character field, because you may want to
enter decimal values (e.g., "1.75", "2.25").
5. MENU REFERENCE 76
-------
Be aware that as the size of text increases, the size of the graph decreases; and as the size of
text decreases, the text becomes less legible. Until you have run the UAMPES once or twice,
we recommend that you use a Size value of "3" for the Title fields and "2" for the Axis label
fields. Note that the Color, Font, and Size specified for the Horizontal Axis label control
the color, font, and size of the bar labels in the bar chart graphic. In addition, for all of the
graphic categories except the spatial graphics category, the Color specified for Title #1 will
also be used to color the box that surrounds the graph and the box that surrounds the legend,
as well as the reference lines and other annotations that appear in the graph. The Color
specified for Title #2 will be used to color the text that appears in the legend.
By allowing the user to completely control the color of the entire display, the user may
eliminate the need for color mapping when sending graphs to a printer. Color mapping is a
process for changing all occurrences of a color (in a single graph) to another color. This
process is a standard feature of the GREPLAY procedure in SAS and is fully described in
Chapter 36 of SAS/GRAPH Software, Version 6 (SAS Institute, Inc., 1990). The use of color
mapping is frequently necessary in a graph that displays the color "white" on the monitor
screen. White shows up quite clearly on the black background of the monitor screen, but
white is invisible on the white background of hard-copy printouts. Color mapping allows the
user to issue an instruction (e.g., to map white to black) when sending a graph to a printer to
make graphic elements visible that would otherwise be invisible. Note that the color mapping
process is usually not necessary if none of the graphic elements are white, and the user can
simplify the printing process by specifying colors other than white.
The graphics category, spatial graphics, is a special case. This category requires a minimum
of eight colors (including the background color) to generate a color graph. However, some
color terminals can only display a maximum of eight colors. Thus, the use of the color white
may be unavoidable. For this reason, certain elements in the spatial graphics (e.g., grid lines,
map lines, and monitor location outlines) are specified as white by the system, and the color
mapping process is unavoidable.
Error Messages
Please enter a size between .
The above message will be displayed if any value other than 0 through 6 is entered in the
Size field. The cursor will be placed on the Size field to allow you to enter a number within
the range of 0 through 6.
There are no error messages for the Text, Color, or Font fields. If an entry in the Color or
Font fields is not valid, the system will select a default entry and plot the graph accordingly.
Note that system defaults may give unexpected results.
77 5. MENU REFERENCE
-------
5.9 AXIS SCALING MENU
The Axis Scaling Menu (2.6) allows users to select the maximum and minimum scale values
for the scaled axes. For example, if all the graphed values are in the range of 500 to 600, the
user may want to specify a minimum scale value of 400 in order to take a closer look at the
details of the graph. Also, if users are creating several graphs to be compared to each other,
this menu allows them to ensure the use of the same scale.
The Axis Scaling Menu is appropriate for only four graphics types: time series, bar chart
(illustrated in Figure 5-18), scatter plot, and quantile plot. (For spatial graphics and bias plot
graphics, the system will proceed directly from the Title/Description Menu to the graphics
display.) For time series and bar chart graphics, only the vertical axis is scaled. Therefore,
for these two graphics, the values chosen in the Axis Scaling Menu will only apply to the
vertical axis. Both the vertical and horizontal axes are scaled in the scatter plots and quantile
plots. By design, both scales are identical; i.e., the maximum, minimum, and interval values
chosen in the Axis Scaling Menu will apply to both axes in the scatter plots and quantile
plots.
r 2.6
UAMPES: BAR CHART AXIS SCALING MENU
Selected
Data Values Axis Values
Maximum 199 200
Minimum 66.26 0
Interval 10
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
Figure 5-18. Bar Chart Axis Scaling Menu.
Instructions
The system checks the data about to be plotted and reports the maximum and minimum
values in the Data Values column of the Axis Scaling Menu. The Selected Axis Values
column provides selection fields for the user to specify maximum and minimum axis values
(which should always encompass the full range of the data values) and the interval to be
indicated along the length of the scale line.
5. MENU REFERENCE 78
-------
In Figure 5-18, a Maximum Selected Axis Value of "200", a Minimum Selected Axis
Value of "0", and an Interval value of "10" have been entered. These values will result in
about 20 labeled tic marks along the scale line.
The fields in the Selected Axis Values column will be blank for the first run of the system.
In subsequent runs, these fields will contain the values from the previous run. Users should
always check the new Data Values against the previous Selected Axis Values to be sure that
the Selected Axis Values encompass the full range of data values. If the Data Values fall
above or below the range of Selected Axis Values, an error message will be displayed (see
discussion below) and no graph will be produced.
At this point, you have given the UAMPES all the information necessary to produce a graph.
After you press , there will be a pause of one or two minutes while the UAMPES
generates the graph you have requested. The graph will be generated and displayed on your
screen. After the graph has been displayed, you may press to continue to the next
menu.
Error Messages
Data Values greater than Maximum Axis Value. . .Reenter.
The above message will be displayed if the Maximum Data Value is larger than the
Maximum Selected Axis Value. The cursor will be placed at the Maximum Selected Axis
Value field to allow you to enter a number that is larger than the Maximum Data Value.
Data Values less than Minimum Axis Value. . .Reenter.
The above message will be displayed if the Minimum Data Value is smaller than the
Minimum Selected Axis Value. The cursor will be placed at the Minimum Selected Axis
Value field to allow you to enter a number that is smaller than the Minimum Data Value.
Range not evenly divisible by step value. . .Reenter.
The above message will be displayed if the difference between the Maximum and Minimum
Selected Axis Values is not evenly divisible by the value entered for the Interval. The
cursor will be placed at the Interval field to allow you to enter a number that will evenly
divide the difference between the Maximum and Minimum Selected Axis Values.
79 5. MENU REFERENCE
-------
5.10 TERMINATION SELECTION MENU
The Termination Selection Menu (2.7), shown in Figure 5-19, presents five possible options
for a variety of actions you may want to take after generating a graphic. The first three
options allow you to restart the system at various points to make changes and/or display the
graphic again. The fourth option allows you to save the graphic in a graphics catalog and
brings up a menu for specifying the options associated with catalog entries. The last option
allows you to exit the system.
-2.7
UAMPES: Termination Selection
Choose the action you would like to take.
1) Change Device or Plot Type
2) Change Input Data Sets or Starting/Ending Conditions
3} Change Titles or Redisplay the Graphic
4) Save the Graphic
5) Exit UAMPES Graphics Module (Without Saving Graphic)
Selection:
Figure 5-19. Termination Selection Menu.
Instructions
Type the number of the option you want in the Selection field and press . Enter " 1"
to return to the Graphic Selection Menu, where you may change any of the options and
generate a completely different graph. Enter "2" to return to the Data File Selection Menu
and generate a similar graph using different data. Enter "3" to return to the Graphic
Title/Description Selection Menu, where you may change the titles and axis labels or make no
changes to simply re-display the graphic. Enter "4" to bring up the Graphics Catalog
Generation Menu, which will allow you to save the graphic that you just generated in a
graphics catalog. Enter "5" to end your current session with the Graphics module and return
to the UAMPES Main Menu.
Error Messages
Invalid entry. Please reenter.
The above message will be displayed if any value other than "1" through "5" is entered in the
Selection field. The cursor will be placed on the Selection field to allow you to enter a
number within the range of 1 through 5.
5. MENU REFERENCE 80
-------
5.11 GRAPHICS CATALOG GENERATION MENU
The Graphics Catalog Generation Menu (2.7.1), shown in Figure 5-20, allows you to
permanently save any or all graphics that you generate in a device-independent SAS Graphics
Catalog.
1-2.7.1
UAMPES: Graphics Catalog Generation
Please enter below the Fully Qualified Data Set Name (DSN)
of the Graphics Catalog to be created or modified.
Do not enclose the Data Set Name in quotes.
Graphics Catalog DSN:
Enter the Name and Description of this graphic. This
information will be stored in the Graphics Catalog and
will aid in identifying the plot.
Name:
Description:
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
Figure 5-20. Graphics Catalog Generation Menu.
Instructions
You will need to provide a data set name (DSN) for the graphics catalog, as well as a name
and brief description of the graph that you have just produced. A single catalog may contain
any number of graphics so you may only need to have a single catalog for all the graphics
you produce. On the other hand, you may wish to have more than one catalog for organiza-
tional purposes.
The Graphics Catalog DSN may be any name that is acceptable to the computer system you
are using. If you are using the EPA's IBM computer system at the NCC, refer to Section 4.5
of this manual for an example and refer to the NCC user's guide (EPA, 1990b) for further
naming conventions.
The UAMPES will display a default Name and Description based on the graphic category
and the first title line of the graph. If the default entries are not satisfactory, you may type
over them with any Name and Description that you prefer. The graphic Name should be no
81 5. MENU REFERENCE
-------
more than 8 alphanumeric characters containing no blanks, and the graphic Description may
be up to 40 characters long and may contain blanks. Refer to Section 4.5 of this manual for
examples.
After you have provided names and a description in the appropriate selection fields, press
to save the graphic to the graphics catalog.
The screen will then display the PROC GREPLAY window; an example is shown in
Section 4.5 (Figure 4-16). PROC GREPLAY will allow you to verify that the graphic has
been placed in the graphics catalog. For complete information on using PROC GREPLAY,
see Chapter 36 of the SAS/GRAPH Software, Version 6 (SAS Institute, Inc., 1990).
Press or to return to the Termination Selection Menu. You may choose to
produce a different graphic by selecting options 1 or 2; take another look at the current
graphic by selecting option 3; save another copy of the current graph (possibly to a second
catalog) by selecting option 4 again; or end the session by selecting option 5.
Error Messages
There are no error messages for the Graphics Catalog Generation Menu. The Graphics
Catalog DSN field is checked by the computer system that you are using. If this data set
name is not valid for the computer system, you will probably see a blank PROC GREPLAY
window without any entry corresponding to the Name and Description fields that you
entered. (The exact response to an invalid data set name is dependent on the computer
system that you are using.) The Name field will be checked by the SAS system. If the
Name that you have entered is not a valid SAS name, the system will enter the default name
(i.e., the graphic name in the Name field before you typed over it).
5. MENU REFERENCE 82
-------
6. FILES
This section describes the formats and contents of files created by the UAMPES.
6.1 FORMAT
Two types of output files are created by the UAMPES: graphics and tables. The graphics
are generated and stored in a binary file using SAS/Graph software and may be replayed or
reproduced on a wide variety of terminals and hard-copy devices using the SAS/Graph
software. If you have access to the UAMPES, you have access to this software.
Tables produced by the UAMPES are saved in TSO data sets as text files. These files are
sequential data sets with a record length of 133 and record format of fixed-block-A. Because
the file is 133 columns wide, you will need to use the and keys to scroll
right and left to see the whole file on your terminal in Interactive System Productivity Facility
(ISPF). If you print these files, be sure that the printer device is formatted to print 132
columns by 60 lines per page. If you need help with viewing or printing, consult the user
support department for the computer system that you are using.
6.2 SAS GRAPHICS CATALOG
The graphics images that you generate with the UAMPES can be saved to a SAS Graphics
Catalog for later use. The topics of generating and saving graphics are discussed in
Sections 4.4, 4.5, and 5.11 of this manual.
Once a graphic is saved in the catalog, it can be replayed by invoking the SAS system at the
TSO READY prompt and using the GREPLAY procedure, or by selecting the Graphics
Catalog option from the UAMPES Main Menu. This option provides a shortcut to the
GREPLAY procedure and substantially reduces the need for knowledge of SAS programming.
The Graphics Catalog option is common to both the UAMPES and the UAMPPS, and is
described in detail in Section 4.3 of the UAMPPS user's guide (EPA, 1990a); see
Appendix D of this manual. The GREPLAY procedure is described in Chapter 36 of
SAS/GRAPH Software, Version 6 (SAS Institute, Inc., 1990). The image can be displayed or
printed on any SAS-supported device, regardless of the device on which it was generated.
For example, if you created a graphic using an IBM 3179G terminal and saved it to a
graphics catalog, you can use the SAS command, PROC GREPLAY, or the Graphics
83 6, FILES
-------
Catalog option to display this graphic on a Tektronix 4211 terminal or any other device
supported by SAS.
For a list of SAS-supported devices, invoke SAS from the TSO READY prompt and submit the
program statement "PROC GDEVICE; RUN;" in the Program Editor window of the SAS
Display Manager. The SAS system will then display a scrollable list of close to 700 devices
that are currently supported on the EPA's IBM system at the NCC.
This list has also been output to a JUSD data set which you may browse and/or print. The
fully qualified data set name is: JUSD.SASGRAPH.V6.DATA(DEVICES). Note that the
parentheses in the data set name indicate that DEVICES is a member in a partitioned data set.
You may find that other members in this data set are very helpful; in particular, members
CAL5845, COM, DEVNOTES, and EXPORT. For assistance, contact the EPA's NCC User
Support Department (see Section 1.4).
6.3 TABLES
For each graph generated by the UAMPES, an accompanying table lists the values that have
been plotted on the graph. For each point plotted on the graph, there is a corresponding value
or pair of values. The titles used in each graph are repeated in the table. In addition, each
table of plotted data values has a footnote at the bottom of each page corresponding to the
label that appears in the lower right hand corner of each graph. Figure 6-1 shows a time-
-series plot of maximum concentrations and its accompanying table. By referring to the titles
and footnotes, there should be no problem determining which table goes with which graph.
6. FILES 84
-------
230
220
210
200-
190
180-
170'
160-
150-
!«
130
120
«°:
100-
90-
80-
70-
60
50-
40-
MAXIMUM 03 VALUES OVER 9 MONITORS
FOR NEW YORK. 11JUL88 HOURS 1 THRU 24
i i
8 «
i i r i i i i i i I i i i i i
10 11 12 13 14 15 18 17 18 19 20 21 22 23 24
METHOD
OBSERVED
BILINEAR
TIME2
MAXIMUM O3 VALUES OVER 9 MONITORS
FOR NEW YORK, 11JUL88 HOURS 1 THRU 24
METHOD* BI LINEAR
HOUR
1
2
3
4
S
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
MAXIMUM
OBSERVED
VALUE
69
77
73
72
63
44
41
46
75
99
125
141
162
206
221
198
199
181
185
154
119
94
79
79
MAXIMUM
PREDICTED
VALUE
105
90
77
67
65
63
66
75
85
98
107
130
140
140
138
147
148
149
124
144
127
124
122
118
PERCENT
DIFFERENCE
-53
-17
-6
7
-3
-44
-62
-62
-14
1
15
7
14
32
38
26
26
16
33
6
-7
-32
-54
-49
Figure 6-1. TIME2 graph and table.
TIME2 DATA
85
6. FILES
-------
In addition to the table of plotted data values, there is a table of summary statistics. This
table provides all seven of the statistical measures for model performance evaluation recom-
mended in Section 5.1.2 of the Guideline for Regulatory Application of the Urban Airshed
Model (EPA, 199la). The table of summary statistics is automatically appended to the table
of plotted data values in the print file specified by the user in the Print File Selection
window.
An example table of summary statistics is illustrated in Figure 6-2. Note the three title lines.
The UAMPES automatically prints the name of the urban airshed region (New York), the date
of the study (11JUL88), the range of hours (1 through 24), the aggregation method (BI-
LINEAR), and the number of monitors (9). The range of hours, aggregation method, and
number of monitors have a direct bearing on the calculation of the statistical measures
provided in the summary statistics table. Because the range of hours, aggregation method,
and/or monitors may change each time a graph is generated, the UAMPES will generate a
summary statistics table for each graph for which an aggregation method is specified (i.e., all
graphics categories except the spatial graphics category). As noted earlier, a summary
statistics table follows each table of plotted data values.
6. FILES 86
-------
Summary Statistics of UAH Recommended Performance Measures
for NEW YORK, 11JUL88, hours 1 through 24
using BILINEAR method and 9 monitors
Perf. Measure
Description
Notation
Units
Value
Unpaired highest
prediction accuracy (1) Au
Normalized
bias test (1) D*
Gross error for all
pairs > 60 ppb Ed*
Average station peak _
prediction accuracy A
Bias of all
pairs > 60 ppb (1) D60
Bias of all
station peaks (1) Dpeak
Fractional Bias for Peak Concentration
Peak Monitor/station Values
ppb
ppb
32.65
-0.95
0.23
25.94
10.84
21.44
Fractional bias
of means (1) Fm
Fractional bias of
standard deviation (1) Fs
Peak Hourly Values
0.32
0.26
Fractional bias
of means (1)
0.06
Fractional bias of
standard deviation (1) Fs
(1) Note: For these measures, a positive value Indicates UNDERpredlction
and a negative value indicates OVERprediction.
Figure 6-2. Summary statistics table.
87
6. FILES
-------
The spatial graphic will produce two different tables of plotted values. The first table lists
cell values used in plotting the map; the second table lists monitor values. These two tables
are illustrated in Figure 6-3. Note that the first two title lines are identical in both tables.
These title lines duplicate the title that will appear on the graph. The third line distinguishes
between the two tables. Note that the table of "Predicted Concentration Levels" lists rows in
descending order. This order holds the rows in the same relative relationship as that seen in
the spatial graphic. That is, the value that is listed in the upper right hand corner of the table
is the same value that is plotted in the upper right hand corner of the spatial graphic. The
table of "Observed Concentration Levels" simply lists the row and column location of each
monitor along with the concentration level.
6. FILES 88
-------
s
p-
p-
UD
p-
p-
rH
p-
o
P-
0
p-
o
p-
rH
p-
S
^
CO
OJ
CO
CO
p-
p-
LO
r-
^D
r-
to
p-
LO
r-
£
p-
CO
u>
CO
CO
rH
CO
O
CO
0
CO
Cn
p-
Cn
p-
s
t-j
cn
O
a\
co
'-O
co
LO
CO
^J*
CO
n
co
CM
CO
s
i.n
cn
m
en
Cn
O
en
Ol
co
en
CO
p~
CO
UD
CO
s
p-
01
LO
en
en
CN
Cn
rH
Cn
rH
en
0
en
en
CO
rH
en
en
CO
CTi
en
in
en
m
en
CN
en
^
en
O
en
f
o
o
^
o
o
o
o
cO
en
u?
en
^
CT>
CN
en
8
i_)
O
rH
oo
o
o
m
o
fN
0
o
o
p-
en
LO
en
S
CO
en
o
o
o
f>
o
m
o
n
o
t_)
o
!H
CO
cn
S
CO
Cn
en
en
o
CN
O
m
O
f*j
O
c^
0
t I
0
0
T-H
s
o
o
o
o
o
t-H
o
^H
o
^
o
^H
o
0
o
s;
en
en
O
0
o
cn
cn
en
cn
cn
cn
cn
cn
cn
cn
S
T-H
o
cn
p-
cn
CO
en
p-
cn
p-
CN
CO
CO
P-
cO
cn
CO
fN
en
^
cn
Lfl
cn
Q E-i
U ft
> a:
w z
(n u
cn u
°g
o
z
o '
o ^ n> o\
ScoS
> W ^
W £K K
J D EH
ce o
MOO
U Cu
D a
r*> ^ Q
O ^H Cd
iH
U
cn
U
CO
U
p-
«>
o
u
CJ
CN
O
in
r-
LT)
r-
L/l
^
CN
LO
CN
LH
CN
LD
CN
i_n
LO
cn
in
cn
p-
LH
ro
LO
CN
LT)
CN
LO
n
LO
^o
^O
rH
cn
r^
Lfl
LD
m
in
ro
i-n
LH
>-O
m
'-U
n
rH
cn
>-o
LO
"^
LO
O
LD
^
LO
<^>
LO
^
W>
^r.
rH
cn
LO
^
LO
LO
m
LO
LO
U3
P-
^
cn
^
^
rH
^
cn
LO
p-
LO
P~
L/l
U3
CO
U3
0
cn
^
f
^D
rH
iJD
en
LO
en
LO
p-
p-
CM
rH
cn
UD
^£)
^
M3
CN
UD
rH
<-o
p-
p-
n
cn
rH
en
vo
^
U)
^
U)
^
U3
P- P-
££
CM 0]
r^ rn
cn -^
p- P-
UJ C^-
U3 CO
UD ^
V£l CO
^O iJD
P-P-P-P-P-P-P-COCO
I^^^^^S^SS
^HrHrM^HCNrfCXIOOCN
OCNCN.^CN.HCN^CO
^.^^^^^^^^
pssssssss
^P?!PP"P^P
cnOiHCNCNCOrnmcn
LOP-P-P-P-P-P-P-P-
S°SS°p^^P^P
CO
CO
LO
r~>
O
p-
p-
S
LO
?
CO
CO
cn
CO
LO
fN
CO
o
CO
CO
S
o
Cu
(fl
D
<
>
O
S
D
S
M
X
s
f£
0
tj
CO
CO
£3
tH
^H
O
z
0
o
Q
^>
K
cn
DQ
O
P
z
z
o
M
EH
Ch
H
CC
O
CO
[d
Q
OOOOOOCNOO
OOOOOOTHOO
OOOOOOCNOO
OOOOOOrHOO
OOO
^%r^
mmn
^HOOOO
^cncnaia\
noooo
CL,
60
00
en
CQ
S
D
I
60
89
6. FILES
-------
The table that accompanies the quantile plot includes a column of deciles in addition to
columns for observed and predicted concentration values. The decile column indicates the
values that were annotated in the quantile plot. In some instances, one or more deciles will
fall between two of the ranked concentrations. If this happens, the system will calculate the
midpoint between the two ranked values. These midpoints will be listed along with the decile
in question, but the rank order of these midpoint values will be indicated by a period "." (see
the 50th percentile shown in Figure 6-4). This figure shows ranked maximum values over
24 hours. The midpoint (i.e., 50th percentile) of 24 hours occurs halfway between the 12th
and 13th ranks. The midpoint values may not appear to be properly calculated due to
rounding. In Figure 6-4, values of 96.5 and 116.5 are rounded up to 97 and 117, respective-
ly. Also, keep in mind that the ranked values listed in the table have also been rounded and
the midpoint is calculated using the ranked values before any rounding has taken place.
MAXIMUM 03 VALUES OVER 9 MONITORS FOR NEW YORK, 11JUL88
(HOURS 1 THROUGH 24)
RANKED RANKED
RANK MAXIMUM MAXIMUM
ORDER OBSERVED PREDICTED
NUMBER VALUES VALUES DECILE
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
221
206
199
198
185
181
162
154
141
125
119
99
97
94
79
79
77
75
73
72
69
63
46
44
41
151
150
146
146
141
138
135
130
126
121
119
118
117
115
106
105
98
93
86
79
75
69
67
67
64
QQ2 DATA
MAX
90%
80%
70%
60%
50%
40%
30%
20%
10%
MIN
Figure 6-4. Quantile plot table.
6. FILES
90
-------
There are two tables that accompany the bias plot (see Figure 6-5).
The first table lists the values from which means and standard deviations are calculated. In
the case of monitor maximums across hours (BIAS1, illustrated in Figure 6-5), this table will
be a list of monitors. The monitors may include all the monitors in the domain or the subset
of monitors selected from the Monitor Selection List Menu. In the case of hourly maximums
across monitors (BIAS2, not illustrated), this table will be a list of all the hours in the range
specified in the Data File Selection Menu. Both BIAS1 and BIAS2 are sensitive to and will
be affected by both the range of hours specified and the monitors selected.
The second table lists (1) the mean and standard deviation of the observed values listed in the
first table, (2) the mean and standard deviation of the predicted values listed in the first table,
and (3) the fractional bias of the two means and the fractional bias of the two standard
deviations.
MAXIMUM 03 VALUES OVER 24 HOURS FOR NEW YORK, 11JUL88
(N = 9 MONITORS - HOURS 1 THROUGH 24)
Source data for calculating means and standard deviations
used in calculating fractional biases.
METHOD= BILINEAR
MONITOR I.D.
MAXIMUM
OBSERVED
VALUE
MAXIMUM
PREDICTED
VALUE
090013007
090031003
090070007
090091123
340170006
340230006
360810004
361030002
361192004
221
185
181
153
199
195
161
172
166.
.00
.00
.00
.00
.00
.00
.00
.00
.00
.74
.26
.17
147.
127.
144.
148.85
111.12
102.03
124.65
140.08
132.79
BIAS1 DATA
MAXIMUM 03 VALUES OVER 24 HOURS FOR NEW YORK, 11JUL88
(N = 9 MONITORS - HOURS 1 THROUGH 24)
METHOD=BILINEAR
MEAN OF STD. DEV. OF MEAN OF STD. DEV. OF FB OF FB OF
OBSERVED VALUE OBSERVED VALUE PREDICTED VALUE PREDICTED VALUE MEAN STD. DEV.
181
21.28
131
BIAS1 DATA
16.38
0.32
0.26
Figure 6-5. Bias plot tables.
91
6. FILES
-------
In certain instances, the tables that accompany two different graphs may be virtually identical
except for the footnote indicating the source of the table. This is the case with the TIME1
and SCAT1 tables illustrated in Figure 6-6. This duplication of data occurs because both
graphs are plotting the same data. The only difference is that the data are plotted in one
graph as a time-series plot and in the other graph as a scatter plot.
A variation of duplicating data may be seen by comparing a TIME1 table with a BAR1 table.
For a given data set, both graphs are based on the same information. The TIME1 graph
compares observed versus predicted values across hours and produces a separate graph for
each monitor. The BAR1 graph compares observed versus predicted values across monitors
and produces a separate graph for each hour. Consequently, the TIME1 table allows a
subheading for each monitor with a group of hours under each monitor. The BAR1 table
allows a subheading for each hour with a group of monitors under each hour. In Figure 6-7,
the values for hour =16 and monitor = 09003 have been highlighted to illustrate how the
same data may simply be tabulated differently.
6. FILES 92
-------
(a)
(b)
03 VALUES FOR NEW YORK, 11JUL88
HOURS 10 THROUGH 18
METHOD = BILINEAR
HOUR
10
11
12
13
14
15
16
17
18
HOUR
10
11
12
13
14
15
16
17
18
COLUMN
18
18
18
18
18
18
18
18
18
COLUMN
23
23
23
23
23
23
23
23
23
MUN1TUK
ROW
13
13
13
13
13
13
13
13
13
MOMTTOR
MUJN 1 1\JK
ROW
22
22
22
22
22
22
22
22
22
1 . JJ. =U2UUI Ji.
OBSERVED
VALUE
71
106
162
206
221
198
148
128
115
T n nQDmic
_L . LJ. UyUUJJA
OBSERVED
VALUE
95
112
101
96
93
109
118
172
185
JU/^^tiiUli
PREDICTED
VALUE
92
98
107
134
140
138
147
148
130
im d.190 1 1
JUJ 4 4 i U 1 1
PREDICTED
VALUE
77
79
71
60
57
60
66
71
75
PERCENT
DIFFERENCE
29
-7
-34
-35
-37
-30
-1
15
13
PERCENT
DIFFERENCE
-19
-30
-29
-38
-39
-45
-44
-59
-59
TIME1 DATA
O3 VALUES FOR NEW YORK, 11JUL88
(ALL MONITORS, HOURS 10 THROUGH 18)
METHOD = BILINEAR
HOUR
10
11
12
13
14
15
16
17
18
HOUR
10
11
12
13
14
15
16
17
18
COLUMN
18
18
18
18
18
18
18
18
18
COLUMN
23
23
23
23
23
23
23
23
23
rakjLM j. ii_it\
ROW
13
13
13
13
13
13
13
13
13
MOM T TO R
MA-JIN ± 1< >t\
ROW
22
22
22
22
22
22
22
22
22
J. . U . -V3 UU J. J*
OBSERVED
VALUE
71
106
162
206
221
198
148
128
115
JVJ / f±"±^U J. -L
PREDICTED
VALUE
92
98
107
134
140
138
147
148
130
PERCENT
DIFFERENCE
29
-7
-34
-35
-37
-30
-1
15
13
I D 090031003442011
OBSERVED PREDICTED
VALUE
95
112
101
96
93
109
118
172
185
VALUE
77
79
71
60
57
60
66
71
75
PERCENT
DIFFERENCE
-19
-30
-29
-38
-39
-45
-44
-59
-59
SCAT1 DATA
Figure 6-6. Comparison of (a) TIME1 and (b) SCAT1 tables.
93
6. FILES
-------
la)
O3 VALUES FOR NEW YORK, 11JUL88
HOURS 10 THROUGH 18
METHOD = BILINEAR
(b)
HOUR
10
11
12
13
14
15
16
17
18
HOUR
10
11
12
13
14
15
16
17
18
iy
COLUMN
18
18
18
18
18
18
18
18
18
>,
COLUMN
23
23
23
23
23
23
23
23 "
23
IUNJ.TUK l.lJ.^UyuulJL
OBSERVED
ROW VALUE
13 71
13 106
13 162
13 206
13 221
13 198
13 148
13 128
-^Ua UVJl _LUUJ^±fi.iUJ._L
OBSERVED PREDICTED
ROW VALUE VALUE
22 95
22 112
22 101
22 96
22 93
22 109
: 22 118
22 172
22 185
TIME1 DATA
O3 VALUES FOR NEW YORK,
MONITOR
I.D.
09001
09Q03
09007
09009
34017
34023
36081
36103
36119
(WITH
COLUMN
18
23
23
20
7
4
10
15
11
PERCENT DIFFERENCE
METHOD" B I LI NEAR HOt
OBSERVED
ROW VALUE
13 148
. 22 118
18 168
15 136
6 199
3 174
7 112
7 121
11 166
77
79
71
60
57
60
...6$
71
75
11JUL88
ANNOTATED)
3&~ 1 6
PREDICTED
VALUE
147
; 66
96
133
110
102
111
104
133
PERCENT
DIFFERENCE
29
-7
-34
-35
-37
-30
-1
15
13
PERCENT
DIFFERENCE
-19
-30
-29
-38
-39
-45
-44
-59
-59
PERCENT
DIFFERENCE
-1
-44
-43
-2
-45
-41
-1
-14
-20
IDTCTTlU/^ri DTT TXTTTTl D U(~M"TD 1*7
MONITOR
I.D.
09001
09003
09007
09009
34017
34023
36081
36103
36119
OBSERVED PREDICTED
COLUMN
18
23
23
20
7
4
10
15
11
ROW VALUE
13 128
22 172
18 181
15 104
6 140
3 138
7 106
7 109
11 135
VALUE
148
71
96
148
111
98
120
104
132
PERCENT
DIFFERENCE
15
-59
-47
42
-21
-29
13
-4
-2
BAR1 DATA
Figure 6-7. Comparison of (a) TIME1 and (b) BAR1 tables.
6, FILES 94
-------
REFERENCES
Cox, W.M., and Tikvart, J.A. 1990. A statistical procedure for determining the best
performing air quality simulation model. Atmospheric Environment, 24A(9):
2387-2395.
EPA. 1990a. User's Guide for the Urban Airshed Model. Volume VI: User's Manual for
the Postprocessing System. U.S. Environmental Protection Agency.
EPA-450-90-007F. Research Triangle Park, NC.
EPA. 1990b. Guide to NCC Services. U.S. Environmental Protection Agency. National
Computer Center. Research Triangle Park, NC.
EPA. 199la. Guideline for Regulatory Application of the Urban Airshed Model. U.S.
Environmental Protection Agency. EPA-450/4-91-013. Research Triangle Park, NC.
EPA. 1991b. Aerometric Information Retrieval System User's Guide. Volumes IV and V.
Office of Air Quality Planning and Standards. National Air Data Branch. U.S.
Environmental Protection Agency. Research Triangle Park, NC.
Press, W.H., Flannery, B.P., Teukolsky, S.A., and Vetterling, W.T. 1986. Numerical
Recipes: The Art of Scientific Computing. Cambridge University Press. New York,
NY.
SAS Institute Inc. 1990. SAS/GRAPH Software: Reference. Version 6, First Edition,
Volume 1. SAS Institute, Incorporated. Gary, NC.
95 7. REFERENCES
-------
APPENDIX A
DATA EXTRACTION
This appendix provides information on the extracted data set, as discussed in Section 1 and
Section 2.2 of this manual. This appendix is identical to Section 4.1 of the user's manual for
the UAM postprocessing system (EPA, 1990a).
A-l
-------
4. SCREEN
4.1 DATA EXTRACTION
The results of a UAM execution contain a large volume of information. All 24 species (see Table 4-2),
a variable number of levels, hourly concentration, and descriptive information about the model run are con-
tained in one file. The Extract module of the UAMPPS provides the data extraction capability for subsetting
the model results based on species and level. In addition, the extraction process transforms the data from a
gridded binary format to either a SAS-formatted data file used in the UAMPPS or a generic text file that can
be imported into other graphics packages. The resulting files are named according to species, level, control
strategy, and date, and thus provide a default method of organizing commonly used data.
TABLE 4-2. DEFINITION OF THE UAM (CB-IV) STATE SPECIES
UAM species
Species name
NO
NO2
O3
OLE
PAR
TOL
XYL
FORM
ALD2
ETH
CRES
MGLY
OPEN
PNA
NXOY
PAN
CO
HONO
H2O2
HNO3
MEOH
ETOH
ISOP
CLBR
Nitric oxide
Nitrogen dioxide
Ozone
Olefinic carbon bond (C=C)
Paraffinic carbon bond (C-C)
Toluene (C6H5-CH3)
Xylene (C6H6-(CH3)2)
Formaldehyde (CH2=O)
High molecular weight aldehydes (RCHO, R > H)
Ethane (CH2=CH2)
Cresol and higher molecular weight phenols
Methyl glyoxal (CH3C(O)C(O)H)
Aromatic ring fragment acid
Peroxynitric acid (HO2NO2)
Total of nitrogen compounds (NO + NO2 + N2O5 + NO3)
Peroxyacyl nitrate (CH3C(O)O2NO2)
Carbon monoxide
Nitrous acid
Hydrogen peroxide
Nitric acid
Methanol (optional)
Ethanol (optional)
Isoprene (optional)
Calibration
A-2
-------
4. SCREEN
Extract: Extract Module Startup Screen
1.0
URBAN AIRSHED MODEL POSTPROCESSING SYSTEM (UAMPPS)
Extract Module
NOTE:
THE EXTRACT MODULE IS CAPABLE OF OUTPUTTING EITHER A SAS FORMAT DATA
FILE FOR PROCESSING BY THE UAMPPS/UAMPES GRAPHICS SYSTEMS OR A GENERIC
TEXT FILE FOR USE WITH OTHER SOFTWARE PROGRAMS.
IF YOU WANT TO PRODUCE THE GENERIC TEXT FILE, YOU MUST ENTER A '₯'
IN THE FIELD BELOW. NOTE THAT SUCH A FILE MAY BE DOWNLOADED AND
MANIPULATED AS WOULD ANY TEXT FILE FOR INPUT INTO OTHER SYSTEMS.
IT MAY NOT, HOWEVER, BE USED AS INPUT FOR UAMPPS OR UAMPES GRAPHICS.
IF YOU DO NOT ENTER A 'Y' THE SYSTEM WILL DEFAULT TO THE SAS FORMAT.
DO YOU WANT TO OUTPUT A GENERIC TEXT FILE (Y/N)? N
PF1/PF13=HELP PF3/PF15=PREV MENU PF4/PF16-MAIN MENU
The Extract Module Startup Screen gives you an opportunity to specify what type of output file will be produced
by the extract. The default output file is an extracted SAS-formatted data set, which is used as input for the
UAMPPS and UAMPES graphics modules. You may choose to output a generic text file by entering a © at
the prompt. If you enter anything other than the © or nothing at all, the SAS-formatted data set will be
produced.
The naming convention for the extracted output file is as follows:
uidacct.UAM.ftyp.EXT.spec.Xddmmmyy.Lnss
where: uid = your User ID supplied by the NCC
acct = your account number supplied by the NCC
ftyp = SASD for SAS-formatted data sets; or TEXT for generic text files
spec = the species you are extracting in this run
dd = the starting day of the UAM run
mmm = the starting month of the UAM run (JAN, FEE, MAR, etc.)
yy = the starting year of the UAM run
n = the number of the level you are extracting in this run
ss = the scenario corresponding to this run. Possibilities are:
1) B (for base case) and the corresponding year, or
2) C (for control) and the corresponding control strategy, i.e., 1,2,3, etc.
A-3
-------
4. SCREEN
Extract: Extract Module Startup Screen
(concluded)
Note that, if you are attempting to perform an extraction for a given species, level, start date combination, and
base/control strategy that has previously been performed, the existing file will be deleted and the results of the
present run will be contained in a data set with the same name.
Instructions:
Press [Enter] to continue. Press [PF4J or (pFi'e]to abort the extract and return to the main menu.
Error Messages:
None.
A-4
-------
4. SCREEN
Extract: File Selection Screen
1,1
UAMPPS: Extract File Selection
Please enter the Fully Qualified Data Set Name for the UAM file from
which you wish to extract a Species and Level subset of information.
Also, supply the name of the Boundary File.
Do not enclose the Data Set Name in quotes.
UAM Output Data Set: uidacct.UAM.MODEL.GR8X8.BS88.JUL1188.AVGCNC
Boundary File Data Set: uidacct.AIRSHED.GR8X8.BS88.JUL1188.BCBIN
Please supply Study Description and Scenario Information
Study: New York Base or Control (B or C):_B Year or Control #: 88.
Please enter the User Id and the Account that will be used to create
the data set name containing the extracted UAM data.
User Id: uid Account Number: acct
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
The information requested on this panel is used to determine the name of the extracted data set and to provide
some information on its name and location.
Instructions:
You must supply fully qualified names of both the UAM output file from which you wish to extract data and
the UAM boundary file. Do NOT enclose the data set name in quotes. Please note that the input file should
be binary and not ASCII or EBCDIC and must be a cataloged disk file residing on the NCC IBM mainframe.
Generally, the name of the domain is placed in the "Study" field, although you may choose any type of identifier.
You must also indicate the Base/Control and the Year/Control indicators in the appropriate fields.
For example, if the file you are performing the extraction on is a base file for the year 1988, then enter "B" in
the "Base or Control" field and "88" in the "Year or Control" field. Otherwise, enter "C" and the control strategy
number in the corresponding fields. Both fields are required. The information you supply is stored as descriptive
information and is used to generate the name of the output data set.
A-5
-------
4. SCREEN
Extract: File Selection Screen
(concluded)
Finally, you must supply User ID and account information in order to determine the prefix of the output data
set being created. These two pieces of information were supplied to you by the NCC when you received your
IBM account authorization.
When you are finished specifying all of the necessary information, press [Enter] to process your selections.
Error Messages:
Entering an invalid file name on either the "UAM Input Data Set" field or the "Boundary File Data Set" field
will result in the following message displayed in the center of the screen:
Press PF15 to continue-
DATA SET DOES NOT EXIST. PLEASE REENTER DATA SET NAME.
If the geographical boundaries in the boundary file do not fall within the geographical domain of the data file,
the following message will appear:
Press PF15 to continue-
FILE DOMAINS ARE INCOMPATIBLE... REENTER ONE OR BOTH.
Press [pFis] to continue and reenter file names.
All other fields are required. If you leave any of the these fields blank, the following message will result:
Press PF15 to continue-
VALUE REQUIRED AT CURSOR LOCATION.
For the "Base or Control" field, any value other than B or C will result in:
-Press PF15 to continue-
PLEASE ENTER B OR C TO CONTINUE.
For the "Year or Control" field, the response must be a numeric value between 0 and 99. Any other response
will result in:
Press PF15 to continue-
VALID RANGE 0-99 ... PLEASE REENTER.
When you press [prislto continue, the cursor will be placed on the appropriate field for data entry.
A-6
-------
4. SCREEN
Extract: Species/Level Data Screen
1.2
UAMPPS: Extract Species/Level Data
Model Run
Start Day: 11JUL88 Hour: 1 End Day: 11JUL88 Hour: 24
Number of Species: 24 Number of Segments: _1
Domain
Columns(x): 31 Rows(y): 25
Vertical Levels(z): 5 ( Upper 3 , Lower 2 )
Enter a ? and Press Enter to Select from a List of Available Species
Select Species: 03 Select Level: 1.
Press PF20 for a Spatial Description of the Domain.
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PFl6=Main Menu
1.2
UAMPPS: Extract Spatial Description of the Domain
UTM Zone: 18
Reference Origin(x): 0 Reference Origin(y): 0
Origin of Grid(x): 520000 Origin of Grid(y): 4460000
Columns(x): 31 Rows(y): 25
Grid Interval(x): 8000 Grid Interval(y): 8000
Vertical Levels(z): 5 ( Upper 3 , Lower 2 )
Height to Base of Surface Level: 0
Minimum Height to Lower Level: 50
Minimum Height to Upper Level: 150
Press PF19 to return to the Species/Level Data Extraction Menu.
PFl/PFl3=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
A-7
-------
4. SCREEN
Extract: Species/Level Data Screen
(concluded)
This screen is a 2-page screen. Page 1 of this screen lists the starting and ending periods of the UAM run and
the species present, along with some spatial information. Page 2 is purely informational, showing the spatial
description of the domain. The number of columns, rows, and vertical levels are repeated from page 1 so that
you will not have to page back and forth to see a complete spatial description of the domain.
Instructions:
On Page 1 of this screen, you are asked to supply the name of the species and the number of the level from
which you would like data extracted. The name of the species must match one of the names supplied in the list
of available species. If you enter a "?" in the "Select Species" field and press [Enter"), a list of available species
will be displayed. You may select a species by placing the cursor on a member of the list and pressing [Enter).
The level you select must be less than or equal to the total number of vertical levels.
When you are finished specifying all the necessary information, press [Enter) to process your selections.
Error Messages:
If the "Select Species" field is left blank or the species entered is not a valid species, the following message is
displayed:
-Press PF15 to continue-
INVALID SPECIES...PLEASE REENTER.
The "Select Level" field must be less than or equal to the number of vertical levels in the domain. If this is not
the case, the following message will be displayed:
Press PF15 to continue-
SELECT LEVEL FIELD OUTSIDE VALID RANGE...REENTER.
When you press
) to continue, the cursor will be placed on the appropriate field for data entry.
A-8
-------
4. SCREEN
Extract: Extraction Process Message Screen
1.3
UAMPPS: EXTRACT DATA FILE
PRESS ENTER TO BEGIN EXTRACTION OF DATA FROM THE UAM FILE
FOR SPECIES 03 AND LEVEL 1
THIS PROCESS MAY TAKE FOUR MINUTES OR LONGER.
THE OUTPUT DATA SET WILL BE CALLED:
'uidacct.UAM.SASD.EXT.03.XHJUL88.LlB88'
THE PREVIOUS COPY OF THIS DATA SET WILL BE DELETED IF IT EXISTS.
*** BE SURE TO MAKE NOTE OF THIS DATA SET NAME FOR LATER USE ***
PFl/PF13=Help PF3/PF15=Prev Menu PF4/PF16=Main Menu
After the Species/Level Data Selection Screen is completed, the extraction process begins. The message screen
is displayed indicating the species and level being extracted along with the name of the extracted data set. At
this time, you should make note of the name of the data set that is being created. When you enter the Graphics
module, you will need to supply this or another extracted data set name. Therefore, it is recommended that
you keep a list containing the names of all data sets you extract along with a brief description of their contents.
After the extraction is complete, you are, of course, free to rename any or all files, if you so desire.
Instructions:
Make note of the name of the extracted data set and be sure to write it down for later use. This data set name
is required by the UAMPPS Graphics module.
Because large data files are involved, the extraction process may take 4 minutes or longer.
Error Messages:
None.
A-9
-------
4. SCREEN
Extract: Termination Selection Screen
1.4
Termination Selection
Select the action you would like to take.
1) Extract Another File
2) Exit the Extraction Module
Selection: 2
The Termination Selection Screen allows you to perform another extraction or to exit the Extract module. If
another extraction is chosen, control is returned to the Extract Module Startup Screen described earlier.
Instructions:
To perform another extraction, enter "1" in the "Selection" field, otherwise, enter "2" to exit the Extract module.
If you choose to exit the Extract module, you will be returned to the UAMPPS main menu.
Error Messages:
The values "1" or "2" are the only valid values for the "Selection" field. Any other entry will result in the following
error message:
Press PF15 to continue-
YOU MUST ENTER 1 OR 2 TO CONTINUE.
Press fFFisho remove the message from the screen. Then, enter a "1" or "2" and press [Enter].
A-10
-------
APPENDIX B
PERFORMANCE MEASURE FORMULATIONS
This appendix provides information on the notation, formulas, and statistical terminology, as
discussed in Section 2.2 of this manual. This appendix is identical in content to Appendix C
of the Guideline Document (EPA, 1991a).
B-l
-------
PERFORMANCE MEASURE FORMULATIONS
RECOMMENDED PERFORMANCE MEASURES1
1. Unpaired Highest-Prediction Accuracy (Au)
=
where
Au = unpaired highest-prediction accuracy
(quantifies the difference between the
magnitude of the highest2 1-hour observed
value and the highest 1-hour predicted value
c0 ( . , . ) = maximum 1-hour observed concentration over
all hours and monitoring stations
cp( . , . ) = maximum 1-hour predicted concentration over
all hours and surface grid squares
1Based on Reference 17.
2"Highest" refers to the maximum 1-hour concentration across
all hours and monitoring stations.
B-2
-------
2. Normalized Bias Test (D
--
where
D* = normalized bias obtained from all hourly
prediction-observation pairs
N = number of monitoring stations
HJ_ = number of hourly prediction-observation pairs
for monitoring station i
NT = total number of station-hours
N
c0(i,j) = observed value at monitoring station i for
hour j
cp(i,j)3 = predicted value at monitoring station i for
hour j
3Predicted value derived from bilinear interpolation of the
predicted values at the four grid cells nearest to station station
i for the given hour.
B-3
-------
3. Gross Error of All Pairs >60 ppb
where
Ed* = normalized gross error for all hourly
prediction-observation pairs for hourly
observed values >60 ppb
NT = total number of station hours (defined
previously)
N = number of monitoring stations
HJ_ = number of hourly prediction-observation pairs
for monitoring station i
c0(i,j ) = observed value >60 ppb at monitoring station
i for hour j
cp (i,j)4 = predicted value at monitoring station i for
hour j
4Predicted value derived from bilinear interpolation of the
predicted values at the four grid cells nearest to station i for
the given hour.
B-4
-------
OTHER SUGGESTED PERFORMANCE MEASURES
1. Average Station Peak Prediction Accuracy A
A = - T |C°(l'f<) cj>(l'V x 100%
where
A = mean paired peak5 prediction accuracies
averaged over all monitoring stations
N = number of monitoring stations
C0(i,ti) = peak observed value at monitoring station i
for hour tj_
Cp(i,ti)6 = predicted value at monitoring station i for
hour ti
"Peak" refers to the daily maximum 1-hour concentration at
a particular monitoring station.
6For these "Other Suggested Performance Measures,
"predicted" can be interpreted in one of several ways: (1) as
the result of bilinear interpolation described in footnote 4 ;
(2) using the procedures described for paired predictions of
daily maxima (described on page 52 of the text); (3) using the
prediction for the grid square containing the monitor site only.
The Modeling Protocol should document the procedure used to
determine "predicted" values in these tests.
B-5
-------
hour of peak observed value at monitoring
station i
2 . Bias of All Pairs >60 ppb(D60)
D - 1 y
60 mr Z4
where
N Hi
Dgg = non-normalized bias from all hourly
prediction-observation pairs for observed
values >60 ppb
NT = total number of station-hours (defined
previously)
N = number of monitoring stations
H| = number of hourly prediction-observation pairs
for monitoring station i
c0(i,j) = observed value >60 ppb at monitoring station
i for hour j
B-6
-------
cp(i,j)7 = predicted value at monitoring station i for
hour j
3. Bias of All Station Peaks (Dpeak)
= ± "
peok \ T £-^ ^ o ^ * i ' p ^ >i'-'
N i=l
where
Dpeak = non-normalized bias from all prediction-
observation pairs for peak8 observed values
at all monitoring stations
N = number of monitoring stations
d/t^ = peak observed value at monitoring station i
for hour t-
7For these "Other Suggested Performance Measures,
"predicted" can be interpreted in one of several ways: (1) as
the result of bilinear interpolation described in footnote 4 ;
(2) using the procedures described for paired predictions of
daily maxima (described on page 52 of the text); (3) using the
prediction for the grid square containing the monitor site only.
The Modeling Protocol should document the procedure used to
determine "predicted" values in these tests.
8 "Peak" refers to the daily maximum 1-hour concentration at
a particular monitoring station.
B-7
-------
cp(i,t.j_)9 = predicted value at monitoring station i for
hour tj_
t.j_ = hour of peak observed value at monitoring
station i
4. Fractional Bias for Peak Concentration
The fractional bias is calculated for both the mean and
standard deviation of peak ozone values, as follows:
Fm = 2x
F .
where
Fm = fractional bias of means
Fs = fractional bias of standard deviation
9For these "Other Suggested Performance Measures,
"predicted" can be interpreted in one of several ways: (1) as
the result of bilinear interpolation described in footnote 4 ;
(2) using the procedures described for paired predictions of
daily maxima (described on page 52 of the text); (3) using the
prediction for the grid square containing the monitor site only.
The Modeling Protocol should document the procedure used to
determine "predicted" values in these tests.
B-8
-------
m0
mean maximum observed concentration
rru = mean peak predicted concentration
s0 = standard deviation of peak observed
concentrations
Sp = standard deviation of peak predicted
concentrations
The means and standard deviations of predicted and observed
concentrations are determined by each of two methods:
Peak station values:
cn{i, .)
maximum observed concentration at monitoring
station i across all hours
10
maximum predicted concentration at monitoring
station i across all hours
where i
1,...,N monitoring stations
Peak hourly values:
10For these "Other Suggested Performance Measures,
"predicted" can be interpreted in one of several ways: (1) as
the result of bilinear interpolation described in footnote 4 ;
(2) using the procedures described for paired predictions of
daily maxima (described on page 52 of the text); (3) using the
prediction for the grid square containing the monitor site only.
The Modeling Protocol should document the procedure used to
determine "predicted" values in these tests.
B-9
-------
c0(.,j) = maximum observed concentration at hour j
across all monitoring stations
Cp(-'li)11 = maximum predicted concentration at hour j
across all monitoring stations
where j = 1,...,H hours
The fractional bias of the mean and standard deviation
varies from -2 to +2. Negative values indicate overprediction
and positive values indicate underprediction.
nFor these "Other Suggested Performance Measures,
"predicted" can be interpreted in one of several ways: (1) as
the result of bilinear interpolation described in footnote 4;
(2) using the procedures described for paired predictions of
daily maxima (described on page 52 of the text); (3) using the
prediction for the grid square containing the monitor site only.
The Modeling Protocol should document the procedure used to
determine "predicted" values in these tests.
B-10
-------
APPENDIX C
ALGORITHMS FOR CELL-AGGREGATION AND FRACTIONAL BIAS
C.I DISTANCE-WEIGHTED ALGORITHM
Both the distance-weighted, nine-cell and the distance-weighted, four-cell aggregation method
use the same algorithm. The algorithm calculates an average concentration that is weighted
according to the squared inverse of the distance from the monitor to the center of each cell.
Therefore, the predicted concentrations of the cell centers that are closer to the monitor are
given greater weight than the values of the cell centers that are farther from the monitor. The
formula for this algorithm is:
* E
D* ^ D2
where PR is the predicted concentration for each cell and D is the distance from cell center to
monitor.
C-l
-------
C.2 FOUR-CELL ALGORITHM
The three aggregation methods that are derived from a square of four cells are: four-cell
closest value, distance-weighted four cell, and bilinear interpolation. These methods use a
common algorithm to select a square of four cells, and then apply individual algorithms to
arrive at a unique aggregation method.
The four-cell algorithm first determines which quadrant of cell 5 the monitor occupies. Next,
it selects the three cells adjacent to that quadrant to complete a square of'four cells. This
process is illustrated in Figure C-l.
If the monitor is in the northwest quadrant of cell 5, then cells 5, 4, 7, and 8 are selected. If
the monitor is in the northeast quadrant of cell 5, then cells 5, 6, 8, and 9 are selected. If the
monitor is in the southwest quadrant of cell 5, then cells 5, 1,2, and 4 are selected. If the
monitor is in the southeast quadrant of cell 5, then cells 5, 2, 3, and 6 are selected.
C-2
-------
(a)
(b)
(c)
(d)
Figure C-l. The square of four cells.
(a) Monitor in northwest quadrant of cell 5.
(b) Monitor in northeast quadrant of cell 5.
(c) Monitor in southwest quadrant of cell 5.
(d) Monitor in southeast quadrant of cell 5.
C-3
-------
C.3 BILINEAR INTERPOLATION ALGORITHM
The bilinear interpolation algorithm calculates a weighted average of the four predicted
concentrations based on the following formulas and illustrated in Figure C-2. Linear
interpolations are calculated along both the x and y axes, hence the term bilinear.
W(Xl,Ym) = W(X1,Y1) + ((Ym - Yl) 4- (Y2 - Yl)) x (W(X1,Y2) - W(X1,Y1))
W(X2,Ym) = W(X2,Y1) + ((Ym - Yl) -r (Y2 - Yl)) x (W(X2,Y2) - W(X2,Y1))
W(Xm,Ym) = W(Xl,Ym) + ((Xm - XI) -f (X2 - XI)) x (W(X2,Ym) - W(Xl,Ym))
where W represents the concentration level at a given point and m represents the monitor
location.
Concentration levels in the formulas correspond to concentration levels in Figure C-2 as
follows:
W(X1,Y1) = Cell 1 concentration level
W(X1,Y2) = Cell 4 concentration level
W(X2,Y1) = Cell 2 concentration level
W(X2,Y2) = Cell 5 concentration level
A linear interpolation is calculated based on the monitor location along the y axis (Ym)
between cells 1 and 4, and between cells 2 and 5. These calculations yield the formula
values of W(Xl,Ym) and W(X2,Ym). A linear interpolation is then calculated between these
two values based on the monitor location along the x axis (Xm). The resulting value
(W(Xm,Ym)) represents the bilinear interpolation.
Note that the cell numbers used in this discussion are specific to the example shown in
Figure C-2, where the monitor is located in the southwest quadrant of cell 5. As the quadrant
location of the monitor changes, the specific cells involved in the interpolation will also
change according to the procedure discussed in Section C.2 of this manual. For a more
thorough discussion of bilinear interpolation, see Chapter 3 in Press, et al. (1986).
C-4
-------
I .
X1
X2
Y2
Ym
Y1
_L'
CX2,YrrO
X1
Xm
X2
Figure C-2. Bilinear interpolation.
[Note: The upper diagram shows a square of four cells; monitor (M) is located in the southwest quadrant
of cell 5; centerlines of selected cells are indicated as XI, X2, Yl, and Y2.
The lower diagram adds reference lines for the *-axis location (Xm) and the _y-axis location (Ym)
of the monitor. The xy coordinates of each point used in the bilinear calculation are given in
parentheses.]
C-5
-------
C.4 FRACTIONAL BIAS ALGORITHM
Bias plots in the UAMPES are specifically designed to plot fractional biases. Fractional
biases generally provide a way of comparing two values and indicating how closely one
matches the other. In the context of the UAMPES, positive values of the fractional bias
indicate that the observed concentration statistic (i.e., mean or standard deviation) is larger
than the predicted concentration statistic. Negative values indicate that the observed concen-
tration statistic is smaller than the predicted statistic. The formula for the fractional bias is
given by:
OB+PR
where OB = statistic calculated using the observed data and PR = statistic calculated using
the predicted data. For a discussion of the fractional bias and its use as a measure of
discrepancy, refer to Cox and Tikvart (1990).
The fractional bias is computed for both the mean and standard deviation of peak ozone
values. The mean and standard deviation of predicted and observed concentrations are
determined by two methods.
(a) The data set corresponding to the peak station values (BIAS1) is determined from the
highest 1-hour average concentrations at each monitoring station as indicated in
Figure C-3(a). For each monitor, both the maximum observed and the maximum
predicted 1-hour average concentrations (from among the hours) are determined. From
these peak station concentration pairs, the fractional bias of the mean and standard
deviation are computed using the fractional bias expression from above. This fractional
bias measures the ability of the model to predict the mean and standard deviation of
observed 1-hour average concentrations from data paired in space but not time.
(b) The data set corresponding to the peak hourly values (BIAS2) is determined from the
highest concentration for each hour as indicated in Figure C-3(b). For each hour, both
the maximum observed and the maximum predicted 1-hour average concentrations (from
among the monitoring stations) are determined. From these peak hourly concentration
pairs, the fractional bias of the mean and standard deviation are computed. This
fractional bias measures the ability of the model to predict the mean and standard
deviation of observed 1-hour average concentrations from data paired in time but not
space.
C-6
-------
(a) Maximum over all hours (BIAS1).
MONITOR
A
B
C
D
E
F
HOUR
12345678
28
32
56
65
41
50
31
24
48
27
39
77
12
$1
'**
42
43
29
59
60
49
34
66
53
72
65
33
53
47
41
46
56
36
70
68
39
63
19
52
79
26
72
56
45
40
38
49
52
Peak Station
Values
'- '^':.':"''';72;*:.
":&. : " 1"*:!!,* "
:'':v'''4'tt::
' '''11: '"'ir
MX,!^
::«. "$»-
"' :7%
Sum(x)=438
Mean = Sum(;r)/W = 438/6 = 73.0
(b) Maximum over all monitors (BIAS2).
MONITOR
A
B
C
D
E
F
Peak Hourly
Values
HOUR
1 2345678
28
32
56
65
41
50
65
31
24
48
27
39
77
;77
12
SI
61
42
43
29
81
59
60
49
34
66
53
66
72
65
33
53
47
41
72
46
56
36
70
68 |
39
68
63
19
52
79
26
72
""791
56
45
40
38
49
52
. ._... .
Sum(x) = 564
Mean = Sum(x)//V = 564/8 = 70.5
Figure C-3. Source of statistics for fractional biases
(applies to observed and predicted concentrations).
C-7
-------
APPENDIX D
GRAPHICS CATALOG
This appendix provides information on the graphics catalog, as discussed in Section 4.5
(Saving the Graphic in a Graphics Catalog) of this manual. This appendix is identical to
Section 4.3 of the user's manual for the UAM postprocessing system (EPA, 1990a).
D-l
-------
4. SCREEN
4.3 GRAPHICS CATALOG
Within the SAS data management system, the Graphics Catalog is the storage area for the graphics
command string that produces graphic output from the processed data. The resulting graphic data can be
"replayed" to redisplay the graphic much more rapidly than repeating the graphic generation process. A
temporary graphics catalog is always produced when SAS graphics procedures are run, but this temporary
catalog and its contents are lost when the SAS session is concluded.
To save graphics output between SAS sessions, you must specify a permanent graphics catalog. The
"Save the Graphic" option on the Termination Selection Screen of the graphics creation module allows you to
create a permanent SAS graphics catalog and to store graphics output in it. To redisplay, merge graphics,
remap colors, or otherwise manipulate the stored graphics, you must run the SAS GREPLAY procedure.
GREPLAY may be run from the TSO Ready prompt by executing the SAS CLIST and subsequently submitting
PROC GREPLAY to the system with the proper parameters. The Graphics Catalog management section of
the UAMPPS allows you a "shortcut" to access PROC GREPLAY from the UAMPPS menu system.
D-2
-------
4. SCREEN
Graphics Catalog: Catalog Management Menu Screen
UAMPPS 3.0
Graphic Catalog Management Menu
Please enter below the Fully Qualified Data Set Name
of the Graphics Catalog to be created or modified.
Do not enclose the Data Set Name in Quotes.
Graphics Catalog DSN: uidacct.UAM.OAOPS.GCATLG
Graphics Device : IBM3179
PFl/PF13=Help PF3/PF15-Prev Menu PF4/PFl6=Main Menu
If you choose the "Graphics Catalog" block on the main menu, you will find yourself in the above screen. This
screen is intended to be a quick way for you to access PROC GREPLAY from within the UAMPPS. You may
create a graphics catalog from this screen or access any existing one.
Instructions:
Enter the fully qualified data set name for the Graphics Catalog. (Do not confuse this data set name with the
extracted UAM data sets.)
Also, enter your graphics device.
When you press (Tenter], the screen will clear and the next screen displayed will be the PROC GREPLAY
screen.
When you have finished using PROC GREPLAY, press [PFIS] to return to the main menu.
Error Messages:
None.
D-3
-------
4. SCREEN
Graphics Catalog: GREPLAY Screen
PROC GREPLAY
Command ===>
IGOUT:
TC:
CC:
GOUT.PLOTS
Sel Name
BAR
BOX
SLINE
CVALUE
CTILE
GOUT:
Template:
Cmap:
Device:
Scroll:
Type Description
J NEW YORK 11JUL88 03 LI B88 GT THRESH
i NEW YORK 11JUL88 03 LI B88 HRLY CONG
I NEW YORK 11JUL88 03 LI B88 HRLY MAXC
1 NEW YORK 11JUL88 03 LI B88 HR OF MAX
I NEW YORK 11JUL88 03 LI B88 PCT DIFF
IBM3179
PAGE
Updated
09/30/91
09/30/91
09/30/91
09/30/91
09/30/91
Field: Description:
"IGOUT" The name of the input catalog (the catalog currently displayed).
"GOUT1 The name of the output catalog (the catalog to which the graphic will be replayed or copied).
"Device" The graphics device on which the graphic will be displayed (generally the current graphics
device).
"TC" The template catalog containing the template to be used when displaying the graphic (leave
blank if no template is used).
"Template" The name of the template to be used to display the graphic (leave blank if no template is
used).
"Scroll" The current scroll value for an entry list greater than one screen in length.
"CC" The color map catalog from which the color map is drawn.
"Cmap" The name of the color map used to display the graphic (leave blank if no color map is used).
"Sel" The selection field; used to enter commands that apply to the corresponding catalog entry.
D-4
-------
4. SCREEN
Graphics Catalog: GREPLAY Screen
(concluded)
Field: Description:
"Name" The name that was assigned to the graphic by the "Name" field on the Graphics Catalog
Generation screen.
"Type" The type of graphics output. See Chapter 3 of SAS/GRAPH Reference for more information.
"Description" The description that was assigned to the graphic by the "Description" field on the Graphics
Catalog Generation screen.
"Updated" The date that the catalog entry was generated.
Instructions:
Display a graphic:
Place an S in the "Sel" field and press [Enter!
Delete a graphic:
Enter DEL in the "Sel" field and press [Enter].
Copy a graphic:
Place a C in the "Sel" field and press [Enter"). The graphic entry will be copied to the catalog specified
in the "GOUT" field.
Display a graphic (or graphics) in a template:
Enter a template catalog name in the "TC" field and the name of the template in the "Template" field.
Next select the graphic (or graphics) from the catalog list by placing an S and a number (or numbers,
i.e. SI, S2, etc.) in the "Sel" field(s) indicating the order in which to display the graphics in the template.
PreSS [Enter].
Change the colors in a graphic:
Enter a color map catalog name in the "CC" field and the name of the color map in the "Cmap" field.
Next select the graphic from the catalog list by placing an S in the selection field. Press (Enter).
Create or edit a template:
Enter a template catalog name in the "TC" field.
On the command line type: edit templatename.TEMPLATE, then press [Enter]. The TEMPLATE
DESIGN window will be displayed for editing of the template. See SAS/GRAPH Reference (pages
1241-1244) for additional information on the TEMPLATE DESIGN window.
Create or edit a color map:
Enter a color map catalog name in the "CC" field.
On the command line type: edit colormapname.CMAP, then press [Enter]. The COLOR MAPPING
window will be displayed for editing of the color map. See SAS/GRAPH Reference (pages 1244-1245)
for additional information on the COLOR MAPPING window.
Change the current catalog:
Enter the new catalog name in the "IGOUT" field and press Center]. The new catalog list will be displayed.
D-5
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1 REPORT NO.
EPA-450/4-90-007G
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
User's Guide for the Urban Airshed Model
Volume VII: User's Manual for the
Performance Evaluation System
5. REPORT DATE
December 19, 1991
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Paul Jarrett (CSC) and Ellen Baldridge (EPA)
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Computer Sciences Corporation (CSC)
Applied Technology Division
P.O. Box 12767
Research Triangle Park, NC 27709
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency (EPA)
Technical Support Division (MD-14)
Office of Air Quality Planning and Standards
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The UAM Performance Evaluation System (UAMPES) is a data display
and analysis tool for evaluating UAM model performance. After
executing the UAM system the results may be graphically compared to
the air quality monitored data. The UAMPES creates time series
plots, tile maps, bar charts, boxplots, scatter plots and quantile
plots. Through a series of menus the user has a wide range of
flexibility in scaling and labeling the graphs. This system
calculates all of the performance measures cited in Appendix C of
the Guideline for Regulatpry Application of the Urban Airshed
Model.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
Urban Airshed Model (UAM)
species, scenario, quantile, tiles,
boxplots, SAS graphics catalog
Aerometric Information Retrieval
System (AIRS), air quality,
fractional bias
18. DISTRIBUTION STATEMENT
19. SECURITY CLASS (This Report/
21. NO. OF PAGES
20. SECURITY CLASS (Tlnspagej
Release Unlimited
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDITION is OBSOLETE
22 PRICE
93
------- |