-------
(on the 1 logarithms of the quantities) to determine a  "true" distribution
of exposures.

Since none of the criteria for employing the analytical solution were
strictly met (observed distributions, although nearly log-normal, failed
the Kolmogorov-Smirov test; errors probably included additive components;
and errors were large enough to cast doubt on the validity of the linear
approximation) a simulation test was developed:  the "true" distribution
was convoluted with the actual errors (drawn with replacement) and the
result compared with the observed distribution.

Results of the calculated corrections and simulations will be presented.
The method appears to be general enough to be employed on other
environmental studies provided sufficient information is available on the
measurement errors.
                                     20

-------
SESSION:  COMPUTER GRAPHICS AND MICROCOMPUTING:  POSTER  SESSION AND
          DEMONSTRATIONS

TITLE:    Statistical Computing:  IBMPC

AUTHOR:   William Smith, SPB, OPPE
Not available.
                                     21

-------
22

-------
SESSION:  COMPUTER GRAPHICS AND MICROCOMPUTING:  POSTER SESSION AND
          DEMONSTRATIONS

TITLE:    Use of Color Graphics to Enhance Your Message

AUTHOR:   Mark Wolcott, QMS, OAR
Just as data presented in the form of black and white graphics are easier
to Interpret than columns of numbers, adding color to your graphs can
highlight and reinforce the message you want to tell.

This presentation provides Insight in how to use color to enhance your
graphics, shows how to get four colors from a three color plotter, and
demonstrates a technique for creating color overlays from a one color
plotter.

Six sample graphs and the Tellagraf commands used to create them will be
provided.
                                     23

-------
24

-------
SESSION:  COMPUTER GRAPHICS AND MICROCOMPUTING:  POSTER SESSION AND
          DEMONSTRATIONS

TITLE:    Microcomputing/Graphics

AUTHOR:   Wilson B. Riggan, HERL, ORD
With the reduction in deaths from contagious disease during  the  past
century, the reduction in deaths due to cardiovascular disease during  the
past decade and a half, and the marked fluctuation in mortaility for
several site-specific cancers during the past 30 years,  the  relative
Importance of geographic patterns of mortaility rates and  trends has
Increased.

Geographic patterns and mapping is not new in epidemiological
investigations.  Dr. John Snow in his classical epidemiologlc study of  1854
used an illustrative map showing the geogrpahic pattern  of the cholera
epidemic in London to generate an hypothesis which he tested.  From these
results, Dr. Snow identified the cause of the epidemic.

Developments in microcomputers, in graphics, and in desk top laser graphic
printers provide us with a new cost effective tool for producing
photocopy-ready illustrative maps of mortaility rates and mortality trends
at the county or state economic area (SEA) level as well as  photocopy-ready
text in support of the maps.  These tools have only recently become
available for use.

Over the three periods covered by this report, there have been marked
fluctuations in mortaility for several site-specific cancers.  Total cancer
mortality rates increased 16% for white males for 39% for nonwhite males.
Among the ten major cancer sites, lung cancer mortality  increased most
noticeably.  White male lung cancer mortality increased  116% from 1950-59
to 1970-79 and white female lung cancer mortality increased  199%.  Nonwhite

                                     25

-------
lung cancer rates rose even more sharply.  Nonwhite male and female  lung
cancer mortality Increased 185% and 188%.  Excluding lung cancer, white
males decreased 5%, white females decreased 15%, and nonwhite females
decreased 13% while nonwhite males increased 15%.

Cancer mortality data as arranged in these maps have many uses.  For
example, the geographic and temporal differences are useful in developing
and examining hypotheses concerning the influence of various environmental
factors.  The degree of consistency of such differences across the four
race-sex groups and among the age groups can furnish additional insights
into possible causative factors.  Elevated rates in even one race-sex group
may suggest hypotheses for further investigation.
                                     26

-------
SESSION:  COMPUTER GRAPHICS AND MICROCOMPUTING:  POSTER  SESSION AND
          DEMONSTRATIONS

TITLE:    Computer Mapping of Air Quality and Emissions  Data

AUTHOR:   George M. Duggan, SASD, OAQPS
The Strategies and Air Standards Division, OAQPS, has developed a  set of
mapping programs for drawing circles (or plotting symbols) and for shading
counties, states, and countries.  These programs are designed for
production work on the Sperry UNIVAC 1100 system located at  the National
Computer Center (NCC), Research Triangle Park, North Carolina.  With this
mapping system, it is possible to generate output for either the CALCOMP
plotter or for previewing on a Tektronix graphics terminal.  Maps  can be
drawn in one or two colors at various sizes from wall display to direct
inclusion in a report.  The circle plotting program can be used to present
data which is specific to some geographic point, e.g., monitoring  site
locations.  Circles whose areas are proportional to observed air quality
values or emission rates have also been drawn.  The spacing or density of
shading is user controlled, thus enabling the user to specify the  same
annotation for each of several maps even when the range of data values is
different on each map.
                                     27

-------
28

-------
SESSION:  COMPUTER GRAPHICS AND MICROCOMPUTING:  POSTER  SESSION AND
          DEMONSTRATIONS

TITLE:    Computer Graphics:  TELAGRAF

AUTHOR:   Barry Bolka, Region V
Computer graphic displays were developed for the Region 5 Environmental
Services Division quarterly report to Headquarters on SPMS commitments.
The TELAGRAF software package on the IBM-NCC mainframe was used to display
the graphs on a Tektronix 4107 terminal, and final copies were produced on
a Houston Instruments DMP-29 multi-pen plotter.

Also, computer maps produced on a Tektronix 4170 PC based system designed
to allow digitizing of new river reach segments for updating to STORET will
be shown.
                                     29

-------
30

-------
SESSION:  COMPUTER GRAPHICS AND MICROCOMPUTING:  POSTER SESSION AND
          DEMONSTRATIONS

TITLE:    The Presentation of Air Quality Data

AUTHOR:   Coe Owen, Region IX
Air quality data may be presented In various ways.  Selected examples will
be shown.
                                     31

-------
32

-------
SBSSIOH:  COMPUTER GRAPHICS AND MICROCOMPUTING:  POSTER  SESSION AND
          DEMONSTRATIONS

TITLE:    National Air Quality and Emissions Trends Report  Graphics

AUTHOR:   Warren P. Freas and Karen A. Nelson, OAQPS, RTP
The National Air Quality and Emissions Trends Report is issued annually by
the Monitoring and Data Analysis Division.  The report is directed  toward
both the technical air pollution audience and the interested general
public.  This presentation focuses on the graphical displays used in  the
Trends Report to depict national and regional trends in air quality and
emissions for each of the criteria pollutants.  Long-term air quality
trends are depicted using boxplots and trend line plots with confidence
intervals.  Additional graphical displays include clustered bar charts for
regional air quality comparisons and shaded component line plots for
national emission trends.  All•graphical displays were prepared using
TELLAGRAF software on the IBM system at the National Computer Center
(NCC).  The boxplot grpahs were developed using a FORTRAN pre-processor,
which operates in a conversational mode, to generate the TELLAGRAF
commands«
                                     33

-------
34

-------
SESSION:  THE USE OF  EPIDEMIOLOGIC DATA IN EPA'S REGULATORY PROCESS
Panel discussion.
27
-  No
                      effect
              U n
                                    35

-------
36

-------
SESSION:  ENVIRONMENTAL EXPOSURE ASSESSMENTS

TITLE:    Statistical Relations Between Exposure and Dosage  for Risk
          Assessments

AUTHOR:   David T. Mage, EMSL, RTF
Risk estimates for exposures to toxic chemicals are often given as a
lifetime risk per average unit dose rate (D, mg/kg-day).  Pollutant
exposures (X, mg/m3) are often measured by sampling of atmospheres indoors
and outdoors, but dosage rate computations require assumptions about the
subjects' inhalation rate (v, liters/min-kg) and fractional deposition
parameter (J = 1 - Xexhaled / Xinhaled).  Physiological variations in v and
J are described, and the effect of the Increase in the variance of the
estimated dosage D - X v J compared to the variance of the exposure X is
discussed.
                                     37

-------
38

-------
SESSION:  ENVIRONMENTAL EXPOSURE ASSESSMENTS

TITLE:    A Mathematical Definition of Human Exposure  and Dose

AUTHOR:   Wayne Ott and Naihua Duan, ORD
Despite the importance of the concepts of exposure and dose  in  the
environmental sciences, no single, detailed definition of  these concepts
has been developed covering all media (air, water, soil, food).  Newly
emerging research studies designed to measure or model man's  exposure  to
environmental pollution require that these concepts be defined  rigorously.
This paper proposes mathematical definitions of dose and exposure that are
suitable for designing measuring programs or developing and  testing
computerized exposure models.  The "target" refers to the  person, place, or
thing being exposed.  The smallest physical setting in which  the  exposure
of a target can occur is a point located on a hypothetical segment of  the
surface of the target (called a contact surface).  The exposure of a point
on this surface is the concentration experienced at an infitesimal point on
that surface at a particular instant of time.  The average exposure over
space Is the integral of the conentratlon over the contact surface, divided
by the area of the contact surface.  Similarly, the average exposure over
time is the integral of the concentration over time, divided  by the
averaging time.  For a dose to occur, a quantity of pollutant must be
transported across the contact surface In some manner, usually  by a
"carrier medium."  Dose accumulation occurs within a envelope made up  of
contact surfaces enclosing the target.  Dose accumulation  is  the  sum of the
integrals over time of the individual doses experienced by all  the contact
surfaces comprising the boundary of the target.
                                     39

-------
40

-------
SESSION:  EXPERIENCE WITH MAJOR STUDIES INVOLVING MULTIPLE ORGANIZATIONAL
          UNITS OVER SEVERAL YEARS

TITLE:    Considerations in Planning Major Surveys

AUTHOR:   Mel Kollander, OPPE
Surveys are playing an increasingly important role in EPA.  In fact, EPA
has become the second largest data collector among regulatory programs.  In
FY'86, two program offices will design nationwide studies costing about 10
million dollars.  This presentation will highlight the factors that should
be considered when developing such large-scale surveys.  A discussion of
contracting, design procedures, interview methods, questionnaire
development, sample design, and analytical procedures will be included.
                                     41

-------
42

-------
SESSION:  EXPERIENCE WITH MAJOR STUDIES INVOLVING MULTIPLE ORGANIZATIONAL
          UNITS OVER SEVERAL YEARS

TITLE:    Experience with the Design and Implementation of the EPA Study of
          Underground Storage Tanks

AUTHOR:   Joseph S. Carra, Chief, and Robert Heath, OTS
The Exposure Evaluation Division of the Office of Toxic Substances is
heading a major, nationwide survey/study to determine whether the problem
of leaking underground tanks is large enough to require Federal regulation,
and to develop sound information upon which to base regulations, if they
prove necessary.  The survey began in February 1984 and will be completed
in October 1985.  A data collection phase, begun in late October 1984, will
be completed in August 1985, and EPA will receive a data analysis by
November 1985.

The survey will provide unbiased national and regional (geographic)
estimates of the number of tanks and leaking tanks, the number of
establishments with tanks, and with leaking tanks, the quantities of motor
fuel leaking into the ground, the rates of leakage, and site-specific data
to develop leak probability models.  These estimates will be used with
national soil inventory maps and corrosivity data bases to estimate leakage
rates and frequencies at the county level.

The survey consists of a national sample of 1,000 randomly selected
establishments containing an estimated 2,600 tanks.  It is being conducted
in the three phases:

     (1)  A questionnaire interview phase.

     (2)  A tank inventory reconciliation phase.

     (3)  A tank tightness-testing phase.
                                     43

-------
The planning and implementation of such a complex survey on this very
controversial subject has been difficult and demanding.  Key problems that
had to be solved included (1) potential difficulties in obtaining adequate
response rates and (2) choosing a tank tightness test with tolerable
measurement errors.
                                     44

-------
SESSION:  EXPERIENCE WITH MAJOR STUDIES INVOLVING MULTIPLE ORGANIZATIONAL
          UNITS OVER SEVERAL TEARS

TITLE:    Attitudes and Experience with Major Studies of Solid Waste
          Problems

AUTHOR:   George Garland, OSW
The time, money, and top management attention necessary for a major survey
incurs reactions ranging from disbelief to outrage.  Nevertheless, major
surveys have taken place in the Office of Solid Waste.  This paper
discusses the magic ingredients to motivate and sustain such efforts.
                                     45

-------
46

-------
SESSION:  EXPERIENCE WITH MAJOR STUDIES INVOLVING MULTIPLE ORGANIZATIONAL
          UNITS OVER SEVERAL YEARS

TITLE:    Review of the Mobile Sources Emissions Factors Survey

AuraoR:   James M. Daley, OSR, OPPE
The Information Management Branch plans to review the survey design and
data collection procedures, and the uses that are made of the data
generated by the Emissions Factors Program.  The Office of Mobile Sources
uses this data for estimating emissions of the Nation's in-use motor
vehicle fleet for evaluating existing and proposed regulations, and for
assessing the attainment of standards.  The study report will recommend
survey management procedures to ensure that this important data collection
activity continues to produce quality estimates.  This paper will describe
study procedures and report study findings.
                                     47

-------
48

-------
SESSION:  STATISTICAL ASPECTS OF AN AMBIENT STANDARD FROM CONCEPTION TO
          IMPLEMENTATION

TITLE:    How Often May Water Quality Criteria Be Exceeded?

AUTHOR:   Maurice E. B. Owens, OWR.S, OW
In our effort to regulate point-source discharges sufficiently to satisfy
water quality criteria in-stream, our program now recognizes that an
In-stream criterion will be exceeded eventually.  This assertion holds
irrespective of how stringently the point source discharge is controlled,
but the level of control does dictate the frequency, duration, and
magnitude of exceedances.  Our Water Quality program has undertaken an
effort to incorporate this stochastic notion into our national criteria
documents and programs for field Implementation.  In debating numerous risk
assessment issues inherent in this effort, a diverse set of proposals arose
on how frequently a criterion may be exceeded.  This presentation is on the
analysis of one.such proposal.  The analysis is formulated to draw together
the proposed statement on frequency, the underlying biological rationale,
and the resulting adverse Impacts relative to an acceptable level of risk
traditionally applied in this area.  Examples with actual stream flow data
are included.
                                     49

-------
50

-------
SESSIOH:  STATISTICAL ASPECTS OF AN AMBIENT STANDARD FROM CONCEPTION TO
          IMPLEMENTATION

TITLE:    Interpreting the Level of Protection Provided by an Ambient
          Standard

AUTHOR:   Thomas B. Feagans, OAQPS, SASD
According to the Clean Air Act, national ambient air quality standards
should protect the public health with an adequate margin of safety.  This
broad and Imprecise guidance suffices for legislative purposes, but there
is a need in setting and reviewing ambient standards to give this guidance
an interpretation that is both more precise and meaningful in the
implementation of the legislative intent.  The paper supports the position
that statistical concepts are essential to such an interpretation.  It
clarifies the sense in which ambient standards provide a level of
protection.  It makes the case that acceptance of several'extant
misconceptions of how an ambient standard provides protection would vitiate
the enterprise of managing envrionmental risks contributed to by air
pollution.  It indicates how compter graphics can enhance this enterprise
by inviting the correct interpretation of the function of ambient
standards.
                                     51

-------
52

-------
SESSION:  STATISTICAL ASPECTS OF AN AMBIENT STANDARD FROM CONCEPTION  TO
          IMPLEMENTATION

TITLE:    Data Analysis Concerns With Respect to  the Form of  the NAAQS

AUTHOR:   Thomas Curran, MDAD, OAQPS
This paper discusses the data analysis considerations that arise in the
development of a National Ambient Air Quality Standard (NAAQS).  This
involves implementation issues such as specifying how ambient data would be
used to determine whether or not the NAAQS has been met.  The choice of a
form and the associated data analysis procedures can affect the relative
stringency of a particular standard so there is an interaction with the
choice of an appropriate concentration level of the standard.

The ozone NAAQS is used to ilustrate how the treatment of these concerns
has changed over time.  Decisions that were made in the development of the
current ozone standard are discussed within the framework of the practical
constraints involved.  Possible changes and enhancements are briefly
considered in the context of future NAAQS revisions.
                                     53

-------
54

-------
SESSION:  STATISTICAL ASPECTS OF AN AMBIENT STANDARD FROM  CONCEPTION  TO
          IMPLEMENTATION

TITLE:    Regional Role in Assessing Compliance With the Ozone  Standard

AUTHOR:   Stephen Goranson, Region V
The Clean Air Act Amendments of 1977, Section 109(d)(i), put  forth
requirements to review the National Ambient Air Quality Standards (NAAQS)
on a regular (5-year) basis.  Review of these standards Includes the
examination of the form of the standard as well as consideration of the
appropriateness of the concentration level and averaging time.  On February
8, 1979, as required under the provisions of Section 108 and  109 of the
amended Clean Air Act, (1) the primary and secondary standards for
photochemical oxidants were changed to 0.12 parts per million (ppm),
designated specifically for ozone, and (2) the definition of  the attainment
point was changed to that point where the expected number of  days per
calendar year with maximum hourly average concentrations above 0.12 ppm is
equal to or less than one.  The new standard provided in a simple form a
more stable planning (target or design) concentration upon which to base
emission reductions, i.e., to develop and verify control strategies.

This discussion will focus specifically on EPA Region V's role In assessing
compliance with.this new form of the ozone standard in some example urban
areas.  The data from these areas typify the range of expected peak ozone
concentrations assuming downwind transport of precursor emissions (volatile
organic compounds and oxides of nitrogen) from center city during the
summer season.  The technique for determining compliance with the standard
and the design values will be exemplified.  Finally, an analysis will be
given of a planned emission control strategy and proposed implementation,
based on the design value for an itnerstate region.  Problems and remaining
issues with such a strategy will be discussed.
                                     55

-------
56

-------
SESSION:  STATISTICAL ASPECTS OF AN AMBIENT STANDARD FROM CONCEPTION TO
          IMPLEMENTATION

TITLE:    State Agency Implementation

AUTHOR:   David Kolaz, Illinois EPA
An ambient air quality standard passes through a variety of stages prior to
Federal promulgation.  If the basic objectives of the standard are Intact
after promulgation, they still face corruption duriag the implementation
phase.  The framework that emerges from promulgation, establishes the
boundaries within which the people involved in the implementation phase
must work.  Illinois' experience in implementing ambient air quality
standards shows that some of the objectives of a standard are actually
impeded by the structure of the standard itself.  Some examples of this
situation are the missing data scheme from the ozone standard, which
purports to "account" for missing data through a simple proportional
multiplier, a statistical form of the ozone standard, which is intended to
reduce the impact of meterological fluctuations, and an averaging scheme
for sulfur dioxide which permits opposing views on the validity of
"running" or "block" averages.

Implementation of an ambient standard also -has connotations other than the
development of control strategies.  Public participation is an important
part of the implementation process and consideration must be given to
providing a clear, consistent basis for describing the quality of the air
to the public.  Presently, the Pollutant Standards Index (PSI) permits an
easily understandable method of conveying air quality information to the
public.  Unfortunately, to the public, the PSI seems somewhat incompatible
with the ozone standard in that it categorizes air as unhealthy when, in
some cases, the standard is attained.  The PSI may also show two full years
of healthy air when the ozone standard Is considered to be violated.
                                     57

-------
These and other concerns with the Implementation of a standard will  be
presented along with some suggestions, from a state perspective, of  how  the
goals of a standard can be most carefully preserved through the entire
process from conception to implementation.
                                     58

-------
SESSIOH:  SESSION ON COMPUTER GRAPHICS AND COMPUTING

TITLE:    A Repord Card on Statistical Software for the PC

AUTHOR:   William P. Smith, SPB, OPPE
These are especially exciting times for the statistician who wishes  to
perform statistical analyses on the microcomputer.  New statistical
packages written for the microcomputer and the recent migration of main
frame stat packages to the micro, now enable statisticians  to attempt
analyses that were once reserved for main frames.  Now we find that  our
analyses can be performed in less time, with greater ease,  lower cost, and
be just as accurate.

The presentation will focus on PC SAS, SPSS PC PLUS, STAT GRAPHICS,  and
SYSTAT.  A summary and comparison of the strengths and special features of
each will be followed by a discussion of their possible roles in
Statistical applications at the EPA.
                                     59

-------
60

-------
SESSION:  APPLICATIONS OF STATISTICS AND COMPUTERS AT EPA

TITLE:    "Your Clients Would Like to Crunch Their Own Numbers?  Why Not
          Design a Custom Course?"

ADTBDR:   Bertram Litt, Office of Pesticide Programs
 1.  Determine Clients Perceived Needs

 2.  Evaluate Strategy to Minimize Misuse of Statistics

 3.  Determine Balance of:

     A)  Traditional
     B)  Controversial
     C)  Developmental Techniques
     D)  Post-Hoc Approaches
     E)  Original R&D

 4.  Level of Discourse to be Used

 5.  Softwear Available and Required

 6.  Preparation Needed for:

     A)  Background
     B)  Critical Topics - Thorough Discussion
     C)  Additional Topics - Limited Discussion
     D)  Additional Nice to Know - If Time Permits

 7.  In-House vs. Contractor Cost Benefits:

     A)  Available Expertise

                                     61

-------
     B)  Production Time and Milestones
     C)  Costs (FTEs/Dollars)

 8.  Briefing and Management Decision

     Assume that:

     A) *Contracting-Out is Selected Path and
     B)  A suitable Contractor is Available

 9.  Schedule Tasks and Deliverables to Meet Milestones:

     A)  Anticipated Differences in Approach Which Will Have to be
         Resolved
     B)  Schedule EPA Decision Points, Review of Drafts, and Interface
         Visits

10.  Regularly Reevaluate Original Plan In Light of Contractor
     Recommendations and Revise Plan as Indicated

11.  Seek and Utilize All Available Support

12.  Pilot Test and Revise the Product
                                     62

-------
SESSION:  APPLICATIONS OF STATISTICS AND COMPUTERS AT EPA

TITLE:    Evaluation of a Laboratory Comparison of a Toxicity Test for
          Drilling Fluids

AUTHOR:   R. Clifton Bailey, Statistics Section, Analysis and Evaluation
          Division, OWRS
The Analysis and Evaluation and the Industrial Technology Divisions in OWRS
and the EPA Gulf Breeze Laboratory worked together to develop a protocol
for a drilling fluid toxicity test to support the NSPS/BAT Effluent
Guideline for the Offshore Segment of the Oil and Gas Extraction Industry.
To satisfy the need for a contractor capable of performing toxicity tests
according to the proposed protocol, the Agency obtained data from 10
laboratories that responded to its Information for Bid Contract
(IFB-WA-W015).  In responding to the IFB, each laboratory applied the
toxicity test protocol to an EPA provided sample of the same well-mixed,
pretested batch of drilling fluid.  The results of the toxicity test for
each laboratory were compared with the results of a concurrent test by
EPA.  The statistical procedures for comparison and evaluation of
interlaboratory differences will be presented.  An interesting feature of
the evaluation was the introduction of concomitant variables into the
standard probit model for bioassay analysis.
                                     63

-------
64

-------
SESSION:  APPLICATIONS OF STATISTICS AND COMPUTERS AT EPA

TITLE:    Mouse - A Computerized Uncertainty Analysis System for
          Environmental Applications

AUTHOR:   Albert J. Klee, Chief, Chemical and Biological Detoxification
          Branch, Hazardous Waste Engineering Research Laboratory,
          Cincinnati
Environmental calculations that involve uncertainties expressed as
probability distributions are far beyond the capabilities of hand analysis
for any but the simplest of models.  There exist a number of computer
simulation languages incorporating Monte Carlo methods that certainly can
do the job, but learning such languages and implementing them on all
computers is often slow and difficult.

MOUSE deals with the problem of uncertainties in models that consist of one
or more algebraic equations which, In turn, contain one or more uncertain
variables.  It was designed to be used by those with little or no knowledge
of computer languages, Monte Carlo simulation, or computer programming.  It
is compact (it runs on everything from mainframe computers down to personal
computers), easy and fast to learn, and has most of the features needed for
substantive uncertainty analysis (built-in probability distributions,
plotting and graphing capabilities, sensitivity analysis, interest
functions for cost analyses, etc.).

MOUSE has been used within EPA in a number of widely different problems
including studying the migration of pollution plumes in streams, designing
a model for the closing of surface impoundment or landfill facilities,
application of engineering fault trees in failure analysis of RCRA land
disposal facilities, cost models for hazardous waste incineration, and the
determination of an appropriate level for regulating organic toxicants in
Industrial wastes discharged to landfills.
                                     65

-------
66

-------
SESSION:  APPLICATIONS OF STATISTICS AND COMPUTERS AT EPA

TITLE:    Stream Segments and Standards (SSS) Data Systems

AUTHOR:   David A. Parrish, Environmental Analysis Section, ESD, Region VI
Regional offices often find themselves dealing with two or more somewhat
incompatible systems.  Those developed by EPA for national purposes and
those developed by state or local agencies for their purposes.  To help to
bridge the gap between these in the area of water quality monitoring data
analysis, Region 6 has developed the "Stream Segments and Standards" or SSS
data system.

State water quality management programs generally focus around their water
quality standards and their water quality planning segments.  State water
quality standards specify numerical criteria for chemical constitutents in
the aquatic environment which are generally more site or area specific
than 'National* standards, such as the old "Red Book."  A State's water
quality planning segments define areas where water quality management
practices are applied.  State segments may or may not correspond to
National systems, such as the EPA's Reach file.  The SSS system is designed
around these two key components of the State water quality program.

Key components of SSS are an SSS number, a segment description, the
beneficial uses, the STORET stations, the USGS flow gages, the Reach
numbers, and other miscellaneous information, such as tidal segments or
scenic rivers.  The SSS number is the system key.  SSS will generate STORET
jobs by segment using stations criteria and parameters from its files.  The
STORET retrievals will include segment descriptions and beneficial uses.
Using the USGS gages flow statistics can be calculated via the STORET flow
file.  The Reach number allows us to combine SSS and STORET data with
information from several files created by OWRS for their effluent
guidelines work to perform a wide variety of analyses.  OWRS files include
data on dischargers, water supplies, and flow parameters.
                                     67

-------
The only problems with SSS have been maintenance of the data, system
support, and programming support.

Despite these problems, we think that the Region 6 Stream Segments and
Standards data system is useful in organizing, retrieving, and analyzing
water quality related data.
                                     68

-------
SESSION:  APPLICATIONS OF STATISTICS AND COMPUTERS AT EPA

TITLE:    An Application of Discriminant Analysis:  State of Wisconsin
          Rivers

AUTHOR:   Arthur Lubin, Region V
Discriminant analysis was applied to the State of Wisconsin rivers to
substantiate that the technique provides a valid mechanism to aggregate
water quality sampling stations.  The analysis tested whether grouping
stations via discriminant analysis is appropriate for stratification.  The
results indicated that though there was a certain level of heterogeneity
within discriminant analysis-defined groups of stations, the
categorizations certainly were adequate.
                                     69

-------
70

-------
WORKSHOPS

-------
                              WORKSHOP SESSION
TITLE:      Part I:    Three-Dimensional Data Graphics
            Part II:  Computing Environments for Interactive Data Analysis

PRESENTER:  Dr. Peter J.  Huber, Harvard University
Part I;  Three-Dimensional Data Graphics

I shall discuss our approaches to, and experiences with, interactive
graphical analysis of high dimensional data.  To Illustrate the issues, a
videotape, "Data Analysis with PRIMH," and a short movie, "Visual Aspects
of 3-d Data Graphics," shall be shown.
Part II;  Computing Environments for Interactive Data Analysis

Most statisticians take computational statistics to mean research on and
with statistical algorithms.  Very few are aware of the complex software
and hardware infra-structure, or computing environment, that is needed if
one is to work effectively wth such algorithms.  A good environment must
support both statistical tactics and statistical strategy:  the former by
aiding with the design, implementation and execution of algorithms
performing specific tasks, and the latter by assisting with the use of such
algorithms in the course of a major data analysis.  I shall discuss
specific requirements those supporting environments should satisfy, and the
challenges involved in building them.  On the tactics side, the desirable
environment is similar to the programming environments used by the AI
community; in addition, numerical mathematics, in particular linear algebra
and minimization libraries must be part of it, as well as computer graphics
of a kind peculiar to statistics.  Many of the requirements on the strategy
side appear to be new and unique to data analysis.  In particular, the
                                     71

-------
environment has to provide very sophisticated retrospective help to the
analyst about an unstructured and potentially messy situation:  once the
environment is able to provide adequate support on the tactics side, the
analyst will overtax his or her own memory by exploring too many branches
in too short a time.
                                     72

-------
                              WORKSHOP SESSION


TITLE:      Business Surveys

PRESENTER:  Thomas B. Jabine, Statistical Consultant
The purpose of the workshops will be to review those parts of the survey
process that require special techniques or are only applicable when the
survey units are businesses, rather than persons or households.  Each topic
will be introduced by the workshop leader, who will then encourage
participants to raise questions or describe their own experiences.

The principal topics to be covered are:

Units of Observation.  Choosing among establishments, employers, companies,
corporations, enterprises, etc.

Frames.  How to get a suitable list to sample from.

Respondents.  How to make sure that the questionnaires go to the right
respondents.

The Standard Industrial Classification.  How to classify units by type of
economic activity.

Pretesting.  Appropriate methods for pretesting business questionnaires.

Special Requirements for Mail Questionnaires.  Most business surveys are
conducted by mail.  What can be done to get more complete and accurate
response?

Sampling from Business Populations.  How to sample from a skewed
population.
                                     73

-------
-{:Wi.li
         CjS

         7
U,!, .,

 /

 (I
                                    U
     |S
                          &>,
                                    p  j
                        OV— :
         ^rr-   r     i
        ^•KffijJi i/vV* :A^
                                     74

-------
                              WORKSHOP SESSION


TITLE:      Spatial Statistics

PRESENTER:  Dr. Paul Switzer, Stanford University
 1.  Statistical Descriptors of Spatial Fields

 2.  Stochastic Models of Spatial Fields and Their Uses

 3.  What is the Practical Meaning of Spatial or Temporal Stationarity
     Hypotheses

 4.  Application to Interpolation and Error Estimation (Kriging)

 5.  Application to Monitoring Network Design

 6.  Multivariate Spatial Fields and Their Reduction

 7.  Temporal-Spatial Fields

 8.  Consolidating Spatial Data for Detection of Temporal Change

 9.  Problems Associated with Monitoring Frequency

10.  The Roles of Variability and Model Error in a Regulatory Setting
                                     75

-------
76

-------

-------

-------

-------

-------