ESTIMATION  METHODS  USED FOR TRI REPORTING
     Laury E. Saligman and Maria I. Menor
Office of Cooperative Environmental Management
     U.S.  Environmental  Protection  Agency,
      1992-1993 NNEMS Fellowship Program
                August  9, 1993

-------
                            Disclaimer
This report was written under the auspices of the National Network
for Environmental Management Studies fellowship (NNEMS) to support
the activities of the National Advisory Council for Environmental
Policy and Technology (NACEPT).   NACEPT is a public advisory
committee established to inform and advise the Administrator and
other officials of the Environmental Protection Agency (EPA) on
policy related issues. This report has been reviewed for approval
by neither NACEPT nor the EPA  and,  hence,, the contents of this
report and recommendations do not necessarily represent the views
of either organization.

-------
                          Acknowledgment: s
The authors would like to thank the following people for their
assistance with this project: David Graham,  Office of Cooperative
Environmental Management; Edward Weiler, Pollution Prevention
Division, Office of Pollution Prevention and Toxics (OPPT); Lisa
Capozzoli, TRI Librarian, OPPT Library.

In addition, the authors would like to thank  Anning Smith, Kathy
Franklin, and John Harman of OPPT for their time and energy spent
reviewing and commenting on the draft version of this report.

-------
                            TAR1.E OF CONTENTS
I.  Introduction	   1
      I.I  Background  on  TRI	  1

      1.2  Report  Background	  4

II. Origins of Basis of Estimate Data  Element	8

III The Study: General Tracking	13
   of Basis of Estimate Reporting on Form R

      III. 1  Purpose  of Research	13

      III.2  Procedure	14
            III.2.1 Charts for Each  Medium	15
            III.2.2   Graphs	15
            IH.2.3 Charts for Each Estimation Method	  16

      III.3  Data  and  Results	16
            Table A: Charts on Number of Estimates
            Table B: Graphs on Number of Estimates
            III.3.1 Fugitive Air Releases	  19
            III.3.2 Point Source Air Releases	   21
            ffl.3.3 Chemical Injected Underground And Released
                       To Water and Land	  22
            III.3.4 Transfers to POTWs and Other Off-Site Locations	  23

      III.4 Charts  for Each Estimation  Method	    24
            III.4.1   Other	  25
            III.4.2 No  Basis  Reported	  26
            III.4.3  Measurements	  28
            III.4.4  Calculations	  29
            III.4.5 Emission Factors	  30

      III.5 Discussion on Data Analysis  	   31

IV. EPA's Data Quality and Enforcement Activities	  32

V. Conclusion  and Recommendations	37

VI.   Bibliography	41

VII.   Appendix	43
      Table: Number of Pounds Released
      Graph: Number of Pounds Released

-------
                         I.   INTRODUCTION


I . 1   BACKGROUND ON  TRI

    In 1986,  Congress enacted the Emergency Planning and Community

Right-to-Know Act, which has two main purposes:  to encourage and

support emergency planning for responding-to chemical accidents;

and to provide local governments and the public with information

about possible chemical hazards in their communities.  One of the

major provisions of this law is toxic chemical release reporting,

which is covered under Section  313 of EPCRA.   Certain businesses

are required to submit  reports each year on the amounts of toxic

chemicals their facilities release into the environment, either

routinely or as a result of accidents. The purpose of this

reporting is to inform government officials and the public of

releases of over 300 toxic chemicals and 20 chemical categories.

These data are maintained and publicly disseminated as the Toxics

Release Inventory (TRI).  TRI regulations require facilities to

report chemical releases and  transfers to the extent that data

are available and reasonably ascertainable,  and to indicate how

they arrived at the reported amounts.

    A facility is subject to the reporting  requirements if all of

the following apply:

    •  It has 10 or more full-time employees.

    •  It conducts manufacturing operations  (that is,  if the
       facility is  included in Standard Industrial Classification
       Codes 20 through 39).

-------
     •   It  manufactures,  processes, or in any other  way uses any of
        the listed toxic chemicals  in amounts greater than the
        threshold quantities.

The   threshold quantities for manufacturers  and processors are as
follows:

     •   75,000  pounds during  the 1987 calendar year
     •   50,000  pounds during the 1988 calendar year
     •   25,000  pounds during the 1989 calendar year  and in
        subsequent years
     •   10,000  pounds otherwise used.1


Each-facility  must complete and file the Toxic Chemical Release

Inventory  Reporting Form (hereafter "Form R") for each listed

chemical or "listed chemical category for which the  facility meets

or exceeds the- thresholds above.

     Items  5 and 6 in Part III of the 1990 Form R  require  the

following  releases and transfers of the chemical  be reported (in

pounds  per year):

Section 5

     •      To  air from fugitive or  non-point sources

     •      To  air from stack or point sources

     •      To  water directly discharged to a receiving stream

     •      In  wastes that are injected underground

Section 6

     •      To water discharged to a- publicly owned treatment works

           (POTW)
•^•Estimating Releases and Waste Treatment Efficiencies For the Toxic Chemical
Release Inventory Form, EPA 560/4-88-002 (Washington, DC:  U.S. Environmental
Protection Agency,  December 1987),  p. 1-1.

-------
           In  other wastes transferred off-site for treatment or
           disposal.*


    After release points  have been  identified,  facilities must

estimate the amount  of toxic chemical released.   Estimates are

based upon monitoring data, mass  balance,  emission factors, or

other approaches such as  engineering calculations and/or judgment

(or some combination of these methods).   The technique used must

be reported on Form  R, and  is referred to as the  Basis of Estimate

for each amount released  to air,  water,  or land.   This reporting

of Basis of Estimate is tied to each type of release or transfer

for each chemical.   It is this data element that  our study focuses

upon.

    Section 313 provides  four categories  on Form  R that submitters

must choose from to  report how they estimate their toxic chemical

releases, and facilities  must enter a letter code that identifies

the method that applies to  the largest  portion of the total

estimated release quantity.  The  codes  are as follows:


    M-  Estimate is based on monitoring data  or measurements  for
        the toxic chemical transferred to  an  off-site  facility.

    C-  Estimate is based on mass-balance  around  entire processes
        or pieces of process equipment. The amount  of  a chemical
        leaving a vessel equals the  amount                entering.
        Therefore,  if input and output or  "product"  streams are
        known  (based on measured values),  a waste  stream can  be
        calculated as the difference between  input  and product  (any
        accumulation/depletion of the chemical in  the  equipment,
        e.g.,  by reaction, must also be accounted  for).
^Estimating Releases and Waste Treatment Efficiencies For the Toxic Chemical
Release Inventory Form, EPA 560/4-88-002 (Washington, DC:  U.S.  Environmental
Protection Agency, December 1987),  p. 2-1.

-------
    E-  Estimate is based on published  emission factors that are
        related to the type of process  or  equipment that the
        facility is using. Such emission factors are those relating
        release quantity to throughput  or  equipment type (air
        emission factors are an example).  Emission factors are
        based on the average measured emissions at several
        facilities in the same industry.

    O-  Estimate is based on other methods such as engineering
        calculations  (for example, estimating  volatilization using
        published mathematical formulas),  non-chemical-specific
        emission factors,or best engineering judgment.!


I.2    REPORT  BACKGROUND

    As the law's title reveals, a main  purpose of the Emergency

Planning and Community Right-to-Know Act is to provide local

governments and the public with information about possible

chemical hazards in their communities.   Citizens have the right to

make their own informed decisions about whether the toxic

chemicals in their communities are a threat to their health and

their environment.  The Toxics Release  Inventory was designed to

help citizens make these decisions,  and the intent of the Basis of

Estimate data element was to provide users with the necessary

information on the origin and quality of the data. What is argued

here is that without a clear understanding of  how TRI release

figures were derived, the decisions  that members of the public

make based on these figures  is that  much less  informed. This study

will evaluate the validity and utility  of  the  "Basis of Estimate"

data element and discuss methods to  increase its usefulness. The

following sections will analyze the  effectiveness of the Basis of
^•Estimating Releases and Waste Treatment Efficiencies For the Toxic Chemical
Release Inventory Form, EPA 560/4-88-002 (Washington, DC:  U.S.  Environmental
Protection Agency,  December 1987), p. 2-6.

-------
Estimate  data element, determine the usefulness  of the information

provided,  and explore the   future of this data element on

subsequent versions of Form R.

    The need  for analysis of  Form R's Basis of Estimate was sparked

by observations by the Office of Pollution Prevention & Toxics'

Pollution Prevention Division staff  (PPD).1  In the  research of

issues dealing with the Toxics  Release Inventory,  PPD had observed

that submitters listed "Other"  as their Basis of  Estimate with a

much higher frequency than  the  remaining estimation  categories.

Toxics in the Community: National and Local Perspectives The 1989

Toxics Release Inventory National Report lists "Other" to be the

Basis of  Estimate" between  17 and 50 percent of the  time,

depending on  the medium.2 This  raises concern since  submitters are

not required  to specify the technique they used when they list

"Other."   For this reason,  the  authors were encouraged to embark

on conducting more detailed analysis into the use  of all four

Basis of  Estimate categories,  focusing on the frequency of

"Other."

     A project  recently conducted by the Research  Triangle

Institute  for PPD was another factor that led to this study.

Titled "Assessment of Changes in Reported TRI Releases Between
1 Ed Weiler of the PPD noticed the preponderance of the use of "Other" in TRI
reporting and suggested that the authors investigated the extent of this
situation.
^  Toxics in the Community: National and Local Perspectives The 1989 Toxics
Release Inventory National Report,  Hampshire Research Associates, Inc.p.128.
This study provides a chart containing the number of pounds released and the
percentage of the total which they represent.  The data analysis in the
present study differs form that in the National report because it examines the
number of estimates made by facilities as opposed to tonnage and provides a
longitudinal comparison over the last for years for number of estimates made
in for releases made to each reporting media.

-------
1989 and 1990," the purpose  of the project was "to assess the

comparative  impact of  real versus paper changes on TRI

submissions.   In particular,  the study focuses on the extent to

which three  factors—changes in measurement  techniques,

production fluctuations,  and source reduction activities—affect

changes in reported TRI  releases."^

    Changes  in measurement technique had the  smallest impact of the

three factors  contributing to changes in TRI  release figures.  24

percent of all facilities surveyed claimed a  reporting change due

to changes in  their measurement or estimation method.  In terms of

quantity of  pounds released,  changes in measurement or estimation

method resulted in an  increase of 82 million  pounds and a decrease

of 104 million pounds  of toxic chemicals reported under TRI

legislation.   The impact of  these changes were considered small

relative to  source reduction and production changes, however, the

authors believe that existence of such paper  change makes the area

worthy of investigation.

    This study will lend further insight to RTI's project and to

EPA's research into the  improvement of Form R.  Part II of this

study will examine the legislative history of the Toxic Release

Inventory in order to  shed light on the decision to include the

Basis of Estimate data element on Form R.  Issues to be discussed

in this section include  the  initial the reasons for including the

Basis of Estimate data element and some of its fundamental
1"Assessment  of Changes in Reported TRI Releases Between 1989 and 1990",
Research  Triangle Institute.  Prepared for U.S. Environmental Protection
Agency, Office of Pollution Prevention & Toxics, Pollution  Prevention
Division, Washington, DC.  EPA Contract No.  CR8818760-01-0 p. 1-1.

-------
limitations.  Part III of the study will analyze what reporting



facilities are doing to ensure that the purpose of this data



element is fulfilled. This section provides information about the



number of instances that each Basis of Estimate is listed for the



release amount of a specific chemical to each release medium.



This analysis is unique in that it compares the first four years



of TRI reporting in terms of estimation technique.  Particular



attention was paid to the "Other" category.



      The data collected here may further supplement the RTI study



which analyzed the results from a sample of reporting facilities



and may lead to further study on the reconfiguration or



elimination of the data element from Form R.  Part IV of this



study  focuses on what EPA  is doing to ensure that the purpose of



the Basis of Estimate data element is fulfilled. Here,  the issues



of compliance and enforcement of TRI's policy regarding the Basis



of Estimate data element are examined.

-------
       II.   ORIGINS  OF  BASIS OF  ESTIMATE   DATA  ELEMENT



    This  section  will discuss the development of  the  Basis of

Estimate  data  element and the decision  to  include it  on Form R.

The research conducted in this respect  included analysis of

relevant  rule-making materials and discussions  with members  of EPA

who were  involved in creating Form R.1

    The designers of Form R decided that users  of TRI data should

have an idea how  the release figures  for each environmental  medium

were derived;  subsequently, the four  categories that  comprise the

Basis of  Estimate data element were created.  According to OPPT's

Environmental  Assistance Division (EAD), the designers of  Form R

did not try to generate a finite number of methodologies that

could be  used  to  estimate release quantities; rather, they devised

estimation categories based on the kind of data reporting

facilities would  have readily available to them.   Therefore, when

TRI was introduced in the Federal Register as a proposed rule in

June 1987, it  was stated that EPA proposed to require a data

element that would indicate whether the quantity  of toxic chemical

reported  was derived primarily based  on monitoring data  for the

wastes leading to release, mass balance calculations  of streams

entering  and leaving process  equipment, emissions factors,  or
^ Interviews  with Larry Longanecker, U.S. Environmental  Protection Agency,
Office of Pollution Prevention and Toxics, Economics,  Exposure and Technology
Division, Chemical Engineering Branch,  Deputy Director;  and
Kathy Franklin, U.S.  Environmental Protection Agency,  Office of Pollution
Prevention and Toxics, Environmental Assistance Division, Toxics Release
Inventory Branch, Chemical Engineer.
                                 8

-------
other approaches,  such as best engineering  judgment.2

Additionally,  the  designers of the data element  hoped that this

data element would be useful for identifying  situations in which

monitoring data  might be obtained as part of  follow-up activities

by EPA or states.

    By December  of the same year,  the Information  Collection

Request for the  Toxic Release Inventory Reporting  Form was sent to

the Office of  Management & Budget, wherein  the purpose for the

Basis of Estimate  data element read as follows:   "...  (to)  provide

users of the data  with a  relative  indication  of  the  quality of

the data and how  it  was  developed."^  According to EAD,  the

emphasis above should be placed on the term "relative" since the

Basis of Estimate  was never meant to be more  than  a  "rough

indicator" of  data quality and how the release figures were

developed.2

    Furthermore, EAD  emphasizes that it would be difficult  for the

data element to  give  a a rough indication of  data  quality.  Data

quality is a complex  issue, being examined  by several  offices

within the agency.  Currently,  there is no  definition  of "data

quality" relating  to  the Basis of Estimate  data  element.   Even if

quality could be defined,  users of the data have no  indication of
^•Federal Register/Vol. 52,  No. 107/ Thursday,  June 4,  1987/ Proposed Rules, p.
21161.

1 From Supporting Statement to Request for OMB Review  of Toxic Chemical
Release Inventory Reporting Form - Title III.  Prepared for Office of Toxic
Substances, December  1987,  p. 10.

^ Discussions with Larry Longanecker and Kathy Franklin.

-------
the percentage error associated with each estimation category

because it would be too difficult to calculate.

    The Basis of Estimate data element is fundamentally limited  in

its ability to provide information regarding the quality of the

release estimates for several other reasons.  Data quality is

linked to accuracy which  depends on the difference between the

actual quantity of chemicals released and the estimated value.

First/ it is impossible to certain of the actual release value.

Second estimated values depend on several factors including:


      •    amount and quantity of  available  monitoring/measurement
          data

      •    the type of estimate necessary if no monitoring  data
          exists

      •    the expertise of the staff completing the report

      •    the nature of the  process involved.1


One conclusion of the site visit program study, then,  and a caveat

for the research presented here is that the "quality" of release
                                             i
estimates in the TRI database will continue to be limited by the

quality and the amount of data available to the individual

reporting facility personnel.

     The Basis of Estimate,  when  entered properly, is more useful

in providing an  indication of how the release estimates were

developed than in providing an indication of the estimates'

quality.  This paper will not address issues of data quality
1 Radian Corporation, "Site Visit Program to Assess 1988 Toxic Release
Inventory Data Quality" July 1991, p.3-28.
                                10

-------
involved with TRI estimates.  First the research presented in this



report will address the extent to which the basis of estimate data



element provides information on how the data was developed.



Without users knowing the how release figures were derived, they



cannot know the quality of the specific measurement or estimate.



The focus of this study will therefore be on the second purpose of



the data element, to inform users of the method used to develop



the data.



    One factor that limits the ability of the data element to



reveal the estimation used in a release figure is the fact that a



single release estimate often involves the use of more than one of



the four estimation techniques from which submitters were required



to choose.  For example, if 40 percent of stack emissions of the



reported  toxic chemical were derived using monitoring data,  30



percent by mass balance, and 30 percent by emission factors,  the



reporting facility would have to enter the letter code "M" for



monitoring.  This letter code does not represent the other two



methods used to find the release estimate and,  therefore, does not



give the full story on how the facility  estimated its stack



emissions of that particular chemical.








EPA allows each facility to decide which estimation method will



provide the most accurate release figure.   To assist facilities,



EPA provides general guidance in choosing an estimation method.



(This guidance will  be discussed in detail in the next section).
                               1 1

-------
    BAD emphasizes that  the Basis of Estimate data e'lement was not

meant to dictate  a preferred hierarchy of estimation methods.

There does  seem to be  an implied preference for the use of

monitoring  data based  on the assumption, that generally it is more

accurate than other  estimation methods.  For example, the guidance

manual,Estimating Releases and Waste Treatment Efficiencies For

the Toxic Chemical Release Inventory Form: Section 313 Emergency

Planning and Community Right-to Know Act of 1986 states "Although

no monitoring is  required to comply with the reporting

requirements, facilities are urged to  use  (readily available)

monitoring  data.ul

    Lastly,  it is important to point  out that throughout the rule-

making process, it was emphasized that reporting this data element

was never meant to create an additional reporting burden on

facilities.  The  Code  of Federal Regulations that implements EPCRA

states:
    "No additional monitoring  or  measurement 'of the quantities or
concentrations of any  toxic chemical released into the
environment, or of the frequency  of such  releases, is required
for  the purpose of  completing this form,  beyond that which is
required under other provisions of law or regulation as part of
routine plant operations."4
1 Estimating Releases and Waste  Treatment Efficiencies For the  Toxic Chemical
Release Inventory Form: Section 313 Emergency Planning and Community Right-to
Know Act of 1986.  p. 1-2
4  40 CFR Ch. 1 (7-1-89 Edition), p. 286.


                                12

-------
              III.    THE  STODY:  GENERAL  TRACKING  OF

             BASIS  OF  ESTIMATE  REPORTING  ON  FORM  R



III.I    PURPOSE OF  RESEARCH

     The  purpose of the following  data  analysis is to evaluate  if

the Basis of  Estimate data element has  successfully fulfilled one

of its original  goals;  to provide  TRI users   with information on

how the data  was developed.   A rigorous  and thorough analysis  of

facilities'  responses will make it possible  to evaluate if the

Basis of Estimate data element provides sufficient information  on

how the release  figures were derived.1

     This study  aims at understanding industry's responses to the

Bases of Estimate.    Although analyzing the  tonnage of chemical

released  leads to important insights on how  the environment and

public are affected,  facilities' behavior is best understood by

analyzing the number of times each estimation technique was used.

The data analysis therefore focuses on  the number of estimates

made by facilities  as opposed to the number  of pounds released

into the environment.   Information pertaining to pounds released

is available  in  the appendix.

     The  data presented in this section provides information on

general patterns of estimation techniques reported from the years
^•As  discussed in section II,  Origins of  Basis of Estimate, the first goal of
the Basis of  Estimate data element as stated on the ICR is to provide users
with a relative indication of  the data quality of release figures.   Data
quality of TRI estimates is too  nebulous  to be captured by the Basis of
Estimate codes given on Form R.  For this reason the data analysis  section of
the study will focus solely on the second purpose of the data element; to
provide information on how data  was developed.
                                13

-------
1987 to 1990.  Analysis over a four year period yields  information

on the distribution of estimation methods reported.  These

responses are intended to  "inform persons about releases of  toxic

chemicals to the environment; to assist governmental agencies,

researchers, and other persons in the conduct of research and data

gathering; to aid in the development of appropriate regulations,

guidelines, and standards; and for other similar purposes."1    The

usefulness of the Basis of Estimate question for TRI users  with

respect to the preceding goals will be the unifying thread

throughout the analysis.

     Special attention will be placed on the "Other" category.

This category, which is inherently vague, can potentially create

confusion in the interpretation of TRI data and impede' users

understanding of the numbers.   The "Other" category will

therefore be rigorously analyzed to help policy makers  transform

the category into one which better fulfills the intentions  of TRI

and the Basis of Estimate question.

III.2    PROCEDURE

     Data was collected using the TRI component file of -the TOXNET

system, an on-line system which accesses the main database-of

filed TRI reports in the National Library of Medicine in Bethesda,

Maryland.  The number of estimates made and the number  of pounds

released or transfered were found for each for the seven -listed

media: non-point air ,  point air releases, water, underground
1TRI Activities Under Section 313 of EPCRA Renewal of Information Collection
Request,  EPA #1363.05, p.2-3


                                14

-------
injection,  land, transfers to publicly owned treatment works, and



transfers to other off-site locations.



III.2.1  CHARTS FOR  EACH  MEDIUM



     Charts were created for each release media and contain the



numbers retrieved from the TRIS system for all four estimation



codes as well as for forms missing a Basis of Estimate code.



These numbers were summed and the percentage of total estimates or



pounds which each estimation technique represents was then



calculated.  These values will be referred to as "percentage



values."  Comparing these percentage values across the four year



period, 1987 through 1990,  allows the observance of any changes



that might  be occurring over time.   That is, a relative decline



in the use  of one method or increase in another becomes clear with



the data displayed in this format.



     Coefficients of variation were calculated for the percentage



values in order to assess their degree of variability over the



four year period.  In order to compare two or more distributions



in terms of their degree of variability,  the most meaningful



measure is  one that captures relative variability.  Defined as the



standard deviation divided by the mean,  the coefficient of



variation measures relative variability,  whereas the standard



deviation measures absolute variability.



III.2.2  GRAPHS



      Graphs were created to provide a visual representation of



the relative use of one estimation method versus another.   Two



sets of graphs were developed;  one set to show the actual  number



of estimates made and the second set to show the percentage of






                               15

-------
total estimates which each estimation method represents.   The



first set of graphs plots the estimation technique on the



horizontal axis and the the number of estimates  (or number of



pounds released) on the vertical axis.   For each estimation



method listed on the horizontal axis there are four vertical



columns corresponding to the four reporting years.  Such a



graphical representation reveals any increase or decrease in a



given estimation method over time.



     The second set of graphs plot the percentage of total



estimates made by each technique against the reporting year.  Each



of the four columns is broken into different color segments which



correspond to the Basis of Estimate code.   This type of



representation.allows comparison of industry's use of the



different estimation codes relative to one another.








III.2.3  CHARTS  FOR  EACH  ESTIMATION  METHOD



     In order to provide a better understanding of the relative



use of each estimation technique for each medium, charts were



created to display the average  percentage value for the four year



period and the associated coefficient of variation.  For each1 of



the four estimation techniques,  the media were listed in



descending order,  beginning with the medium which the designated



estimation method was most frequently used to determine TRI



releases.



111.3  DATA  AND  RESULTS



     In order to make the data analysis treatable, information  is



presented in four sections,  each pertaining one of the following





                               16

-------
release  media  :  Section  III.3.1 fugitive air, Section III.3.2



point source air, Section  III.3.3. underground injections, land,



and water, and  (4) transfers to publicly owned treatment works,



and  other off-site locations.  These charts show the percentage



of estimates made by each  given estimation technique for the



reporting years 1987 through 1990 inclusive.



     For each medium or group of media the charts are arranged as



follows.  The left hand column identifies the estimation Basis of



Estimate code indicated by facilities on Form R.   The four



estimation codes provided by Form R   (i.e. Measurements,



Calculations, Emission Factors, and "Other") as well as two



additional categories were incorporated into the chart.    A "No



Basis Reported" category was included because in some instances no



code was entered.  A "Multiple Categories" variable was added



because releases to 1)  water,  and 2)  land, as well as transfers



to both 1) POTWS and 2) other off-site locations are divided into



several subcategories of release media.  Part III, Section 5.3 of



the 1990 Form R requires separate responses for estimates of



discharges to each distinct receiving stream or body of water;



Part III,  Section 5.5 requires facilities to list separately



releases to on-site landfills,  land treatment/ application



farming, surface impoundment, and other disposal;  and  Part III,



Section 6 requires facilities to report transfers to each POTW and



other off-site location separately.   Estimates made for more than



one subcategory under these media are grouped together under



"multiple subcategories reported".   The set-up of TRIS makes



pooling the numbers together  the most useful and efficient  means






                               17

-------
of data analysis.  Furthermore,  the number of "Multiple



Categories" entries represent a  very small percentage of total



records (between 0% and 4% for percentage  of estimates)  and are



therefore not expected to have a significant effect on the study's



results.
                               18

-------
                                 Number of Estimates Made
       Fugitive Air  Releases
40.000TT
35,000
30,000
25,0001-
      Stack Air Releases
        Releases  to Water
10,000-
 6,000
 6,000
Releases  Injected Underground

-------

7,000
6,000-
5.ooo-f
4.000-f
1
3,000-f
2,000-f
i,ooo4
o4






Releases to Land








fflte™


i
B
•
•
P

. 	






*> 0 « ^ "
I | 1 i I i
= •= f 
-------
                  Percentage of Estimates Recorded with Each Basis of Estimate Code
         Fugitive Air  Releases
        /Vps
     /A
iwK^mlm
 90%
 80%
 70%-
 60%-
 50%-
 40%-
 30%-
 20%-
 10%
  0%
       1987
               1088    1989
                              1990
                                                                Stack Air Releases
                                                                                   D No Basis
                                                                                     Reported
                                                                                   (13 Otter
                                                             1987
                                                                     1988    1989
                                                                                     1990
          Releases to Water
100%
 80%
 60% 1
 40%
 20%i
       1987    1988    1989    1990
                                                           Releases Injected Underground
                                                      100%
                                                       80%
                                                       60%
                                                       40%
                                                       20%-
                                                             1987     1888     1989    1990

-------
        Releases to Land
    1087    1088    1088    1080
Transfers  to  Off-Site  Locations
   1087
           1088
1080
                            1800
                                  Transfers to  Publicly  Ownec
                                                   Works
   No Basis Reported

   Other

   Emissions Factor

   Calculations

LJ Measurements
                                          1087    1088   1080
                                                                 1000

-------
III.3.1
FUGITIVE  AIR  RELEASES
Fugitive Air Releases
Percentage of
Measurements
Percentage of Mass
Balance Calculations
Percentage of
Emission Factors
Percentage of "Other"
Percentage of No
Basis Reported
1987
4
13
15
62
6
1988
4
14
13
63
6
1989
4
14
13
63
6
1990
4
13
13
63
6
Average
4
14
14
63
6
CV
0.02
0.02
0.06
0.01
0.02
      Generally,  engineering estimates and other approaches  are

the most applicable methods to estimate fugitive air emissions.1

These estimates are recorded in the  "Other" category.  Fugitive

emissions are usually diffuse and therefore difficult to directly

measure.^   Similarly,  mass balance  calculations tend not to be

the most applicable estimation method  because the release from a

leaking source is  small relative to  the total quantity of

chemicals and therefore can be easily  masked by inaccuracies in

input and output values. ^

     The "Other"  category encompasses  several different yet

frequently used estimation methods for fugitive emissions.  For

example, in certain cases emission factors  are the most applicable

estimation method,  however many available emission factors are not
IA slide from EPA's Form R Training Course Environmental Assistance Division
of OPPTS
o
   EstJ.matj.ng Releases and Waste Treatment Efficiencies For the Toxic Chemical
Release Inventory Form, EPA 560/4-88-002  (Washington, DC:  U.S. Environmental
Protection Agency,  December 1987), p.3-17
3 Ibid, 2-8
                                19

-------
chemical specific but for VOCs  or pariculates which then  require

adjustments.   Such estimates are  coded in the "Other" category and

not  in  the "Emission Factors" category.1

      In addition, computer  modeling programs are sometimes  used in

determining chemical releases from volatilization of on-site

treatment of aqueous or solid wastes.  Knowledge of the major

pathways for the chemical,  type of waste  site,  and set of

meteorological conditions must  be accounted for in the estimate.

EPA  has created computer models for estimating secondary  releases

from nonaerated impoundments, aerated  impoundments, disposal

impoundments,  land treatment, and landfills.  Estimates made using

these models are also reported  with the Basis of Estimate code

marked  as."Other."2
•'•Kathy Franklin, Comments on Draft Report,  6/7/93.
^Estimating Releases and Waste Treatment Efficiencies For the Toxic Chemical
Release Inventory Form, -EPA 560/4-88-002 (Washington, DC:  U.S. Environmental
.Protection Agency, December 1987), p.3-r20


                                20

-------
     The overwhelming majority  (i.e. 63 percent) of estimates are



recorded in the "Other" category.  When TRI users review the



database, it is impossible for them to know if the majority  of



the estimates of fugitive release figures were developed from



engineering estimates,  non-chemical specific emission factors,



EPA modeling.programs, or other methods entirely.  The release



figures derived from all these different methods are coded in the



"Other" category.
III.3.2
POINT  SOURCE  AIR  RELEASES

Percentage of
Measurements
Percentage of Mass
Balance Calculations
Percentage of
Emission Factors
Percentage of Other
Percentage of No
Basis Reported
1987
7
13
9
50
22
1988
8
15
10
61
6
1989
8
15
10
61
6
1990
8
15
10
61
6
i
Average
8
15
10
58
10
cv
0.08
0.08
0.06
0.10
0.82
     "A relatively greater percentage of point source air emission



estimates than fugitive air emissions were measured directly  (8%



for the former as compared to 4% for the latter).   However, 91% of



the point source releases are not monitored directly revealing



that facilities rely upon  mass balance calculations,  emissions



factors',  and other techniques to estimate point source releases.



Data show mass balance calculations to be a major  method of



estimating point source air releases (15% of total estimates on
                               21

-------
the average for the four year period), then followed by emission

factors (10% )  and then measurements  (8%).   Again, the "Other"

category dominates the field, representing 58% of the point source

estimates made during the four year period.


III.4.3    CHEMICALS  INJECTED UNDERGROUND  AND  RELEASED  TO
WATER  AND  LAND
Underground
Injections
Percentage of M
Percentage of C
Percentage of E
Percentage of O
Percentage of NBR
Releases to Water
Percentage of M
Percentage of C
Percentage of E
Percentage of O
Percentage of NBR
Percentage Multiples
Releases to Land
Percentage of M
Percentage of C
Percentage of E
Percentage of O
Percentage of NBR
Percentage Multiples
1987
11
8
2
76
3
1987
28
6
2
51
12
1
1987
11
8
2
65
12
2
1988
16
6
2
76
1
1988
34
4
2
55
3
1
1988
13
7
3
71
5
2
1989
18
7
2
71
1
1989
36
4
3
54
2
1
1989
12
7
3
72
4
1
1990
18
7
2
63
9
1990
36
4
2
53
3
1
1990
12
6
3
70
7
1
Average
16
7
2
71
3
Average
33
5
2
53
5
1
Average
12
7
3
70
7
2
CV
0.20
0.15
0.07
0.09
1.14
CV
0.12
0.24
0.15
0.03
0.89
0.01
CV
0.06
0.09
0.12
0.05
0.54
0.15
     It is interesting to point out the consistency of choice of

estimation technique used for estimates to these media.  As seen

by a low coefficient of variation, the percentage of estimates

made using each estimation technique remained nearly constant.

However,  for all three media the percentage of forms submitted
                               22

-------
without a Basis of Estimate dropped slightly from the  first



reporting year to the second as the percentage of estimates



placed in the "Other" category increased.  This phenomena  will  be



discussed in the section III.5,  Data Arranged by Estimation



Method: No Basis Reported.
III.3.4
TRANSFERS  TO  POTWS  AND  OTHER  OFF-SITE  LOCATIONS
Transfers to POTWs
Percentage of M
Percentage of C
Percentage of E
Percentage of O
Percentage of NBR
Percentage of Multiples
Transfers to Other Off-
Site Locations
Percentage of M
Percentage of C
Percentage of E
Percentage of O
Percentage of NBR
Percentage of Multiples
1987
26
11
1
44
18
0
1987
22
14
1
38
22
3
1988
33
12
1
51
3
0
1988
25
17
1
50
4
3
1989
36
10
1
50
2
0
1989
24
17
1
51
3
4
1990
39
9
1
48
3
0
1990
24
16
1
52
3
1 4
Average
34
10
1
48
7
0
Average
24
16
1
47
8
4
CV
0.17
0.10
0.10
0.06
1.11
0.68
CV
0.06
0.09
0.13
0.14
1.18
0.06
     Transfers, both to publicly owned treatment works  and other




off-site locations, were reported under measurements more  than  any




of the other media. Off-site transfers (other than POTWs)  tend  to




be weighed for total amount of waste due to shipping and RCRA




manifesting policies.  POTWs usually require facilities to monitor




wastewater.  Furthermore, these discharges  often have permitting




requirements.1
l-Kathy Franklin.  Comments on Draft Report. 6-7-93
                               23

-------
     During the four year period at least 34% of transfers to



POTWs and 25% of transfers to other off-site locations were



directly measured.  These figures far exceed the number of



estimates made by  the methods of mass balance calculations (17%)



and emission factors (1%).  Similar to point source emissions,  and



releases to water and land,  the number of forms submitted without



a Basis of Estimate code decreased from 1987 to 1988,  and then



remained fairly constant.   However,  the drop was much more



pronounced for transfers which plummeted from 18% in 1987 to 3% in



1988 for transfers to POTWs,  and fell from 22% to 4% for transfers



to other off-site locations.








III.4    CHARTS FOR EACH  ESTIMATION METHOD



     For each estimation method, the data was organized by Basis



of Estimate to gain a better understanding of facilities' tendency



to employ each estimation technique.   Five charts were produced



which correspond to the four estimation codes (Measurements,  Mass



Balance Calculations,  Emission Factors,  and Other)  as  well as the



no basis reported category.   The release media  were listed in



descending order,  beginning  with the  medium which the  designated



estimation method was most frequently used to determine TRI



releases.
                               24

-------
III.4.1  OTHER
          Percentage  of  Estimates  Coded  as  "Other"
Releases
Underground Injection
Land
Fugitive Air
Point Source Air
Water
1987
76
65
62
50
51
1988
76
71
63
61
55
1989
71
72
63
61
54
1990
63
70
63
61
53
Average
72
70
63
58
53
Coefficien't
of
Variation
0.09
0.04
0.01
0.09
0.03

Transfers
Other Locations
P.O.T.W.s

38
44

50
51

51
50

52
48

48
48

0.14
0.06
     The overwhelming tendency for facilities to report  TRI



releases in the other category is clearly displayed in the



preceding chart.   Depending on the medium,  the average number of



estimates recorded in the "other category"  was between 48% and



72%.  Around seven out of ten estimates made for underground



injection and releases to land were coded as "Other".    Six out of



ten estimates to air (i.e. fugitive and point source)  were coded



as "Other".  Transfers,  to POTWs and to other off-site locations,



represented the smallest percentage of use  of the other category



which in both cases was  around 48%.



     The broadness of the "Other" category  impedes user's



understanding of the how the data element was developed and



therefore prevents the Basis of Estimate data element  from



fulfilling its original  purpose.   Rather than providing users
                               25

-------
information on how the data was developed,  it conceals from them



the origins of the TRI release figures.     The "Other" category



includes estimates made by engineering judgments,  non-chemical



specific engineering calculations,  computer modeling programs,  and



unpublished emission factors.   It is therefore impossible for



users to know how the majority of TRI estimates were made.



     Furthermore, because of the "Other" category,  users are



unaware of not only which estimation technique was  employed,  but



if facilities were switching from one method to another within  the



"Other" category during the four year period.  Perhaps one  year a



facility used best engineering judgment  and the following year



SOCMI emission factors.  Using only publicly available data,  users



would be totally unaware of this change.
III.4.2  No  Basis  Reported
              Percentage  of   No  Basis  Reported
KsDeasffiS
Point Source Air
Land
Fugitive Air
Water
Underground Injection
1987
22
12
6
12
3
1988
6
5
6
3
1
1989
6
4
6
2
1
1990
6
7
6
1
9
Average
10
7
6
5
4
Coefficient
of
Variation
0.80
0.51
0.00
1.13
1.08

Trams IF® ITS
Other Locations
P.O.T.W.s

22
18
4
3
3
2
3
3
8
7
1.18
1.12
                               26

-------
     For all media except  fugitive air and underground  injections,



the number of forms submitted without a Basis of Estimate code



dropped significantly between the first reporting year,,1987, and



subsequent years.   One explanation for this drop is that EPA



detected the missing information and notified facilities to



correct the forms. x  It  is also important  to point out that



following this initial drop, the percentage of forms without a



corresponding Basis of Element remained fairly constant.   Users



therefore did not gain more knowledge of the derivation of data



over time.



     Furthermore, the  decrease in the numbers of forms submitted



without a Basis of Estimate code did not necessarily  improve



users understanding of how the data was developed.  For most



media,  a decrease in "No Basis Reported" from 1987 to 1988



coincided with an increase of estimates coded as "Other."



Although  the instant drop in forms lacking a Basis of Estimate



code showed that facilities gained a better understanding of Form



R and TRI reporting procedures, TRI users did not benefit from



this learning process.  They remained unaware of how the release



figures were derived.
^Kathy Franklin.  Comments on Draft Report.  6-7-93






                               27

-------
111.4.3  MEASUREMENTS
        Percentage  of  Estimates  made  by  Measurements
EffiDefflSffiS
Water
Underground Injection
Land
Point Source Air
Fugitive Air
1987
28
11
11
7
4
1988
34
16
13
8
4
1989
36
18
12
8
4
1990
36
18
12
8
4
Average
34
16
12
8
4
Coefficient
of
Variation
0.12
0.20
0.06
0.08
0.02

Trannsffffirs
P.O.T.W.s
Other Locations
26
22
33
25
36
24
39
24
34
24
0.17
0.06
     For most  of the media the percentage of estimates made by

direct measurements  increased 1987 to  1988  and then stabilized. A

greater percentage  of point source release estimates were made

using direct measurements than for fugitive  air emissions.  Point-

source air emissions are confined and therefore more likely to

have been measured  than random and diffuse fugitive emissions.^
^Estimating Releases and Waste  Treatment Efficiencies For the Toxic Chemical
Release Inventory Form, EPA 560/4-88-002  (Washington, DC:  U.S.  Environmental
Protection Agency,  December 1987), p.  2-8
                                28

-------
       Percentage  of Estimates  made  by Mass Balance
                         Calculations
K
-------
III.4.5  EMISSION  FACTORS
   Percentage  of  Estimates  Made  Using   Emission Factors
K©D©ffl§®§
Fugitive Air
Point Source
Land
Water
Underground Injection

TT[rsinn§ff(er§
P.O.T.W.
Other Locations
1987
15
9
2
2
2


1
1
1988
13
10
3
2
2


1
1
1989
13
10
3
2
2
1990
- 13
10
3
3
2
Average
14
10
3
2
2
Coefficient
of
Variation
0.06
0.06
0.02
0.15
0.07


1
1

1
1

1
1

0.10
0.11
     Chemical specific emission factors remained a consistently



small percentage of total estimates.   This  technique  was employed



most frequently for releases to air(14% for fugitive  and 10%  for



stack emissions).   For all other media the  use  of chemical



specific emission factors was almost  negligible (1% to  3%). These



figures remained consistently low for all four  reporting years.
                               30

-------
III. 5  DISCUSSION ON DATA ANALYSIS



     This analysis used information on the public database to show



that the Basis of Estimate data element does not adequately



fulfill its original goal of informing users about the derivation



of reported release figures.   Using only publicly available



information,  it is impossible for users to have a clear



understanding of how TRI data was developed.  If the Agency still



believes" this goal to be important, the data element must be



restructured.



    The greatest obstacle to user's understanding of the derivation



of release figures is the magnitude and broadness of the "Other"



category.   The majority of estimates for all releases and



transfers  were reported with Basis of Estimate marked as "Other".



The broadness of this conceals from users any indication of how



the data was developed and thus prevents the data element from



fulfilling its original purpose.
                               31

-------
    IV.    EPA'S  DATA  QUALITY  AND  ENFORCEMENT  ACTIVITIES



    It is EPA's responsibility to guarantee  that the information

provided through  the Basis  of Estimates  is conveyed to the public

as completely as  possible.   This  section will discuss what EPA is

doing to guarantee that  TRI  users   have  a relative indication of

the quality of the data  and  how  it  was developed.   From the data

generated for this study, it appears  that the Basis of Estimate

data element is not adequately tracked through TRI's data quality

process, since a  percentage  of submitters fails to comply with

this reporting requirement every  year.   This question can be best

answered by detailing the data management procedure. After

industry submits  Form Rs to  EPA's EPCRA  Reporting  Center, these

forms are processed, entered into a Local Area  Network  (LAN)

database  and a sample  is verified  for data  accuracy.  Records

from the LAN database are periodically uploaded to EPA's mainframe

computer.  EPCRA  Reporting Center staff,  with the  assistance of

EPA personnel, use terminals connected to the  mainframe to run

data quality reports, analyze the data for accuracy, and

standardize data  (such   as parent company names) across the

database.1

     Submitting facilities are contacted, when necessary, to

notify them of noncompliance with  the regulation  and to resolve

data quality errors.  Notices of Noncompliance (NONs) are sent to
•'•Toxic Release Inventory Productivity Review, Office of Toxic Substances, U.S.
Environmental Protection Agency,  July, 1990, p. 8.
                                32

-------
facilities that  have failed to provide or meet  a critical

requirement  such as the reporting year or signature.  Such errors

impede data  input to the mainframe base; failure to report a Basis

of Estimate, therefore, would not result in a NON.1  Failure to

comply with  this reporting requirement, however,  may result in the

mailing of a Notices of Technical Error  (NOTE),  which identifies

erroneous or inconsistent  data submitted to EPA and states.2

NOTEs_are generated using the mainframe database and are triggered

by three types of technical errors:  Categories I,  II,  and III.

Category I is comprised of the technical errors that are

considered the most serious; Category II errors are considered

major, and Category III errors are considered minor by

comparison.* One Category I error or two Category II errors on

one Form-R triggers the NOTE process  (i.e. sending a NOTE to a

facility, which  is described below.

    Before they  are mailed to facilities, NOTES are reviewed,  first

by a.team of technical experts who are contractors such as the

Science Applications International Corporation;  this team

determines whether the error conditions identified in the NOTEs

are valid or whether corrections can be made, with the submitter's

approval, by the contractor.  NOTEs contain two parts:   1)  the
       p. 54.

^Discussion with Kim Orr, Office of Pollution Prevention and Toxic  Substances,
Office of Pollution Prevention and Toxics, TRI Information Management Branch,
Information Management Division, TRI Data Administration Section.

* April 23, 1993 discussion with Kathy Franklin, U.S. Environmental Protection
Agency,.Office of Pollution Prevention and Toxics, Environmental Assistance
Division, Toxics Release Inventory Branch, Chemical Engineer.
                                33

-------
 "Notice," which  is  a standard letter  explaining what the  NOTE  is

 and the requirements for responding to it, and  2) an error report

 that is specific to the section of the form in which errors are

 identified, explaining to the submitter how to correct the

 error(s).  Failure  by a submitter to include a Basis of Estimate

 is a Category  II error; therefore,  this and another Category II

 error would result  in the submitter receiving a NOTE.

 Unfortunately, budget limitations have prohibited the enforcement

 of this process;  therefore,  EPA has relied on a policy of

 voluntary response  on the part of submitters receiving NOTEs.**

    Permeating TRI's  data management procedure is an overall data

 quality program  that  has the following objectives:  "1) :   identify

 and assist reporting facilities so that the data  submitted will

 be of the highest quality;  2)   insure high quality data entry; 3)

 correct and normalize as much of the data as possible in order to

 maximize the utility  of the  data;  and 4)  accurately assess the

 relative validity of  release estimates and other data."1  While

 the first three  objectives are not  directly relevant to the Basis

 of Estimate's original intent to provide information regarding the

 relative quality (as  well as derivation)  of release estimates, the

 fourth objective  speaks directly to this purpose.   Folded into

this objective is the goal to continue conducting site visits to

evaluate the accuracy of emissions  estimates.    EPA last
** ibid.
1-1990 Toxics Release Inventory Public Data Release, Office of Pollution
Prevention and Toxics, EPA 700-S-92-002 (Washington, DC:  U.S. Environmental
Protection Agency, May 1992), pp.  198-199.
                                34

-------
contracted Radian to conduct site visits in order to  determine  how

well facilities used available data to estimate  1987  and  1988

releases.

    The four categories that make up the Basis of Estimate data

element have never been reviewed by EPA.   As mentioned earlier in

this report, it is difficult to use the Basis of Estimate as a

relative indicator of the quality of the release estimate because

it is difficult to pinpoint the accuracy of each estimation method

as they relate to the release figure.  The site visits were

designed to reveal, among other things, whether more  accurate

release estimation  methods could have been used, based on

available information to the facilities.  It turns  out that the

site surveyors agreed with the calculation method used by the

facility for approximately 84% of release estimates.1  What is

more significant for discussion is the two-pronged approach site

surveyors used for identifying errors in release estimates.

    First,  the surveyors always recalculated releases using the

same technical approach used by the facility:  Second, whenever

the-surveyors' experience and training indicated that a

calculation approach different from that used by the  facility

personnel was appropriate, the surveyor attempted to  obtain .the

data needed to calculate releases using the more  appropriate

approach. (In fact, in the site visit report, accuracy was

measured by how closely facility estimates compared to the site

surveyors'  estimates.)   Further, though the four release
l"Site Visit Program to Assess 1988 Toxic Release Inventory Data Quality" July
1991, p.3-28.


                               35

-------
estimation techniques  described in Section 313 and currently

available to reporting facilities are monitoring data, mass

balance, emission  factors,  and other (engineering calculations),

surveyors used what they  considered to be a "more detailed

approach"  in reviewing release estimation techniques which

considered the following:

     1)  Monitoring   data or direct measurements;

     2)  Published emission factors;

     3)  Information  from hazardous waste manifests;

     4)  Mass balance  calculations;

     5)  Engineering calculations;

     6)  Engineering  judgment; and

     7)  Other techniques including facility -derived emission

         factors.x

     In addition to continuing its site visit programs, EPA

intends to continue accuracy evaluations of TRI data by making

phone calls to facilities  that fail certain edit checks for

quality of the technical  data submitted on TRI's Form R.  EPA has

also developed a guidance  manual for EPA Regional  inspectors on

what to look for when  auditing an  EPCRA reporting facility.  The

manual provides inspectors  with detailed guidance on "how to

determine if a facility has identified all reportable chemicals

and calculated thresholds  properly and if the releases seem

reasonable."2
     Toxics Release Inventory Public Data Release, Office of Pollution
Prevention and Toxics, EPA 700-S-92-002 (Washington, DC:  U.S. Environmental
Protection Agency, May 1992), p. 199.


                                36

-------
              VI  CONCLUSION  AND  RECOMMENDATIONS


     This report has shown that the Basis of Estimate data element

in its current form does not successfully provide TRI users with

knowledge of  (1)  the release figures'  data  quality and (2)  how

the data was developed.  This is due mainly  to the

disproportionately large size of the "Other" category and the

fundamental limitations of the data element.  Furthermore,

although TRI has a process that helps guarantee that information

submitted on Form R is complete, budget constraints have

prohibited EPA from providing adequate  enforcement measures.   In

spite of these limitations,  EPA can improve  the usefulness of the

information provided by TRI  through the following recommendations.

The  authors  offer  recommendations  (1) and  (2)  as  a two-
step  process to  help  EPA   determine  the  future  of the  data
element:


(1)  EPA should conduct outreach efforts to  discover the extent to
which users are interested in the data  quality and  derivation of
TRI release figures.

     If such a study reveals that users do not find this

information useful the Basis of Estimate is  needlessly burdensome

and should be deleted.   In specific,  industry wastes time in

answering the question,  and  EPA wastes  time  and valuable computer

space by entering the responses into TRIS.


(2.1) If users require information on how release figures are
derived, the Basis of Estimate must be  redesigned to fulfill this
need.
                               37

-------
      • Additional Basis of  Estimate Codes should be  given to
common estimation techniques.

      The  research undertaken  by Radian Corporation during the 1990

and 1991  site  visit programs  demonstrate that it is  possible to

create estimation categories  that  are more detailed  than  those

which exist  today.   The first  step in reconfiguring  this  data

element should be to determine  the various estimation  methods

which facilities  have coded in the the "Other" category.   If there

are only  one or two different   techniques, then these techniques

should be given their own codes.   If several methods, or

combination  of methods comprise this data element, then EPA must

develop a means of  representing these different estimation

techniques-'-.

      Furthermore  users should be aware of the cases  in  which

several estimation  techniques  are  used for one reported release.

This objective could be obtained by (1)  reporting all the relevant

estimation codes  or (2) placing a  secondary code next to  the Basis

of Estimate  code  which notifies users that several estimation

techniques were used for a single  estimate.


      • Require submitters to  specify estimation method  when
      listing "Other" as Basis of Estimate.
1 For example, engineering calculations and best engineering judgment are
acknowledged and discussed in the  ICR and the TRI reporting package
instructions.  If these types of estimates are so prevalent and considered
equally acceptable as mass balance calculations and emission factors, then the
agency  should identify them to TRI users by coding them on the TRI data base.
Clearly engineering calculations and best engineering judgment are very vague
terms, but by incorporating them into Form R and then the TRI data base, users
would have a better indication of  how the data was developed.


                                38

-------
     It there are a multitude of different estimation techniques

which comprise the "Other" category, it may be too difficult to

represent all estimation methods by additional codes. Although

this information will not be available on on-line, it could made

publicly available through contacting the  EPCRA reporting center.

However, the first alternative is preferable because the

information would be more accessible from the on-line database and

therefore is more in keeping with the spirit of EPCRA.



(2.2) Should users value having an indication of release figures'
data quality, a means of obtaining this information must be
developed.


     The concept of data quality of estimates is too complicated

to be captured through a self-reported Basis of Estimate data

responses.  EPA has not and cannot attach accuracy ratings to each

estimation method.    An alternative method to asses data  quality

must be developed, implemented,   and conveyed to TRI users.   One

possibility to determine the data quality is 'to conduct facility

audits similar to those originally intended by TRI's data quality

program.


4)  EPA should provide increased guidance to submitters who fail
to comply with the Basis of Estimate reporting requirement.

     An analysis of EPA's data management procedure for TRI data

shows that the failure to report a Basis of Estimate for a release

figure can trigger sending a NOTE to a facility.   Unfortunately,

budget constraints have limited EPA in its efforts to follow up on

the NOTEs process with adequate enforcement.   EPA has reported
                               39

-------
some success with the current policy of voluntary response.



However,  the Agency must decide how important it is for submitters



of the data to comply with all reporting requirements of TRI and



should increase guidance,to improve compliance.
                              40

-------
                           VI.  SOURCES


Booz, Allen & Hamilton. 1990. Toxic Release Inventory Productivity
Review,  prepared for Office of Toxic Substances, U.S.
Environmental Protection Agency, Washington,.DC.

Environmental Protection Agency.  1989.  40 Code of Federal
Regulations,  Part 372-Toxic Chemical Release Reporting: Community
Right-To-Know.   (7-1-89 Edition).

Federal Register/ Vol. 52, No. 107/ Thursday,  June 4, 1987/
Proposed Rules.

Franklin, Kathy.  U.S. Environmental Protection Agency, Office of
Pollution Prevention and Toxics, Environmental -Assistance
Division, Toxics Release Inventory Branch, Chemical Engineer.

Hampshire Research  Associates, Inc.  1991. Toxics In The
Community: National and Local Perspectives.    Prepared for  U.S.
Environmental Protection Agency,Office of Toxic Substances,
Economics and Technology Division, Washington, D.C.

Longanecker, Larry.  U.S. Environmental Protection Agency, Office
of Pollution Prevention and Toxics, Economics, Exposure and
Technology Division, Chemical Engineering Branch, Deputy Director.
April 23, 1993 discussion.

Office of Pesticides and Toxics.  1987.  Estimating Releases and
Waste Treatment Efficiencies For the Toxic Chemical Release
Inventory Form,  Washington, DC:  U.S. Environmental Protection
Agency, December 1987.  EPA 560/4-88-002

Office of Prevention, Pesticides and Toxic Substances,Office of Pollution
Prevention and Toxics.  1992. 1950 Toxics Release Inventory Public
Data Release,  Washington, DC:  U.S. Environmental Protection
Agency.  May 1992.   EPA 700-S-92-002

Office of Toxic Substances.  "Supporting Statement to Request for
OMB Review of Toxic Chemical Release Inventory Reporting Form -
Title III."  Prepared for Office of Toxic Substances,  Washington,
DC.  December 1987.

 Office of Toxic Substances.   1991. Toxic Chemical Release
Inventory Reporting Package for 1990, EPA 560-4-91-001
(Washington, DC: U.S. Environmental Protection Agency).

Orr, Kim. Office of Prevention, Pesticides and Toxic Substances,
Office of Pollution Prevention and Toxic Substances,  Office of
Pollution Prevention and Toxics, TRI Information Management
Branch, Information Management Division, TRI Data Administration
Section.
                               41

-------
Radian Cooperation. 1991. "Site Visit Program to Assess 1988 Toxic
Release Inventory Data Quality,"  Radian Corporation.  Prepared
for U.S. Environmental Protection Agency, Office of Toxic
Substances, Washington, DC. July 1991.

Research Triangle Institute.  1993.  "Assessment of Changes in
Reported TRI Releases Between 1989 and 1990".  Prepared for U.S.
Environmental Protection Agency, Office of Prevention, Pesticides
and Toxic Substances,  Office of Pollution Prevention & Toxics,
Pollution Prevention Division, Washington, DC.  EPA Contract No.
CR8818760-01-0.

 Weiler, Edward,  Office of Prevention, Pesticides and Toxic
Substances, Office of Pollution Prevention & Toxics, Pollution
Prevention Division, Washington, DC
                               42

-------
                          VIZ   APPENDIX



The charts and graphs contained in this section refer to tonnage



of TRI chemicals released to the environment.  Table VILA



displays the number of pounds released to each media as recorded




under each estimation technique for all four reporting years.  The



chart is divided into seven subsections which correspond to the



seven media.  Each subsection contains the number  of total pounds



recorded by each estimation technique for all reporting facilities



from 1987 to 1990 inclusive.  The left hand column identifies the



the bases of estimate which facilities can indicate on form R.



Measurements, calculations, emission factors, and "Other" are the



categories provided by form R.



     Graph VILA is a set of seven graphs which show the actual



number of pounds released into the environment for each medium.




These graphs reveal an increase or decrease in the actual amount



of chemicals released into the environment.  Graph VII.B shows the



percentage of pounds released as determined by each estimation




technique.  These charts indicate a rise and-fall in the in the



relative use of one estimation technique as compared to another as



marked by a change in tonnage.
                               43

-------
Table VII. A:  Pounds  Released to Each Medium

F. for
Measurements
Calculations
| 1987 | 1988 | 1969 | 1990

33.068.263 31.746.369 28.737.948 26.603.654
268.730.978 268.106.143 264.532.951 220.063.321
Emission Factors 119.632.010 120.943.198 93.SS0.975 89.062.831
Other
372.523.171 378.813.168 35S.998.1S4 324.969.221
No Basis Reported . ^ ,^- ;. 	 •Ł. -Ł&A*™ijiZjZ..-^^.'; %^J^ .. s^ ,-, •> '. -'~
246.325,782 297.SS8.70S 282.363.923 274.372.337
601.539.217 606.535.442 581.385.856 530.714.538
Emission Factors 131.S99.73S 136.473.236 116.775.989 107.347.720
Other
639.680.142 763.215.289 707.987.428 602.665.247
No Basis Reported (NBR) 298.S63.O68 77.317.843 64.539.894 43.029.556

Total

1.917.707.944 1.681.100.521 1.753.053.092 1.556.129.398
Percentage of M 13 16 16 18
Percentage of C 31 32 33 34
Percentage of E 7 7 7 7
Percentage of 0 33 41 40 39
Percentage of NBR 16 4 4 3
Ircentage 100 1OO 1OO 1OO
Measurements
Calculations
*"" "*" 1.279J61SJ652 "l 5?'. 187.1567' 100.548.658 151.368.120
4.639.737.417 46.118.073 24.15S.1S9 10.434.839
Emission Factors 130.389.641 7.324.407 1.679.647 1.617.427
Other
2.493.421.910 52.040.200 56.862.530 31.732.142
NO Basis Reported 677.199.237 68.778.202 2.649.160 1.422.165
Multiple Categories 51.670.478 3.875.092 2.659.699 776.496

Total
Percentage of M
Percentage of C
Percentage of E
Percentage of O

9.272.034.335 335.323.541 188.755.053 197.351.189
14 47 S3 77
50 14 13 S
1211
27 16 30 16
Percentage of NBR 7 21 2 1
Percentage Multiples 1110
Total Percentage 100 100 100 1OO
Flje.CfwTOBgg^giig ^j
Measurements
Calculations

729.803.992 672.773.853 352.116.234 297.724.691
1.679.650.736 94.547.349 34.284.720 36.323.101
Emission Factors 470.760 601.273 825.575 1.466.572
Other
499.695.001 596.653.998 778.150.477 387.082.916
No Basis Reported 172,273.084 530.439 2.897.577 15.779.967

Total
P mage of M
ige of C
jge of E
PelCKitage of O

3.081.893.595 1.365.106.912 1.168.276.583 738.379.247
24 49 30 40
55 7 3 5
O 0 0 O
16 44 67 52
Percentage of NBR 6 O 0 2
Total Percentage 100 100 100 100







Ave. (or Gil
four years
4
33
14
46
•
,






Ave. for all
lour years
16
33
7
38
7







Ave. lor an
lour years
48
20
1
22
6








Ave. for an
our years
36
17
0
45
2

Coefficient of
Variation
O.O4
O.03
0.10
O.02
0.43







Coefficient of
Variation
O.I3
0.04
O.04
O.O9
0.93







Coefficient of
Variation
0.54
0.98
0.48
0.34
1.22








Coefficient of
Variation
0.31
1.43
O.S9
0.47
1.26

                    Pagel

-------

Land
Measurements
Calculations
Emission Factors
Other
No Basis Reported (NBR)
Multiple Categories

Total
Percentage of M
Percentage of C
Percentage of E
Percentage of O
Percentage NBR
Percentage of Multiples
Total Percentage
Patwt
Measurements
Calculations
Emission Factors
Other
No Basis Reported
Multiple Categories

Total
Percentage of M
Percentage of C
Percentage of E
Percentage of O
Percentage of NBR
Multiple Categories
Total Percentage
Olfc?«te ./Transfers^..,.....
Measurements
Calculations
Emission Factors
Other
No Basis Reported
Multiple Categories

Total
Percentage of M
Percentage of C
Percentage of E
'ercentage of O
Percentage of NBR
Percentage of Multiples
Total Percentage
| 1987 |

SS2.438.119
794.494.116
9.393.258
683.123.700
286.606.412
37.030.867

2.363.086.472
23
34
0
29
12
2
100

	 336>86.423
407.61 7.753
86.561.817
659.956.272
404.640.374
0

1.895.562.639
18
22
5
35
21
0
100

577.065.914
354.434.723
10.457.044
661.879.423
395.258.325
92.771.084

2.091.866.513
28
17
0
32
19
4
100
1988 |

311.570.410
308.586.847
12.001.431
418.179.891
33.292.860
75.889.058

1.159.520.497
27
27
1
36
3
7
100

166.426.787
252.637.729
21.482.130
262.809.299
26.739.740
80.571

730.176.256
23
35
3
36
4
0
100
	
654.053.377
455.320.519
9.025.115
714.647.871
51.018.279
79.701.404

1.963. 766.565
33
23
0
36
3
4
100
1989 |

141.502.341
36.934.332
4.133.731
175.494.996
21.690.083
91.710.547

471.466.030
30
8
1
37
5
19
100

141.581.346
185.170.757
29.038.579
189.826.059
8.140.001
4.248

553.760.990
26
33
5
34
1
0
100

361. 760.430""
168.882.058
9.307.951
318.258.518
20.140.919
55.282.184

933.632.060
39
16
1
34
2
6
100
1990

216.057.982
21.983.410
1.302.941
156.104.083
25.336.858
28.271.093

449.056.367
48
<
0
35
6
6
100

113.429.652
139.549.641
32.316.316
161.209.640
2.837.550
523.624

449.866.423
25
31
7
36
1
0
100
•w
364.219.624
128.608.779
13.335.450
265.214.395
11.902.636
38.982.275

822.263.159
44
16
2
32
1
5
100








Ave for all
four years
32
18
1
34
6
8








Ave. for all
four years
23
30
5
35
7
0







i
Ave. for all
four years
36
18
1
34
6
5









Coefficient of
Variation
0.34
0.77
0.56
0.11
0.64
0.91








Coefficient of
Variation
0.16
0.20
0.35
0.02
1.45
1.76








Coefficient of
Variation
0.20
0.18
0.61
0.06
1.34
0.17

Page 2

-------
                                          Table   VII. A
                            Number of Pounds Released to the Environment
          Fugitive Air  Releases
 400,000.000-
 350.000.000^
 300,000.000-"
 250.000,000-
 200,000.000-
 150,000,000-
 100.000,000
  50.000,000^
          0
             Stack Air Releases
  800,000.000
  700,000.000
  600.000.000
  500,000.000-
  400.000.000
  300,000.000
  200,000.000
  100.000.000-
           0
           Releases  to Water
5,000.000.000
4.500.000,000-
4,000,000,000-
3.500,000,000^
3,000,000,000-
2,500.000,000
2,000,000,000 '
1,500,000,000-;
1,000,000.000-1
  500,000,000-'
           0
      Releases  Injected  Underground
1,800.000,000-
1,800,000,000-
1.400,000,000 -
1,200,000,000
1,000,000,000-
 800.000.000-
 600,000,000
 400.000.000
 200.000.000
           0

-------
          Releases to Land
800,000,000-
700,000,000-
600,000.000-
500.000,000^
400,000,000
300,000,000-^
200,000.000-r
100.000.000
         0
    Transfers to  Off-Site  Locations
800.000.000
700,000,000-
600,000.000
500.000,000
400.000,000-"
300,000,000-
200.000.000
100.000.000
         0
                                                               Transfers  to Public Owned
                                                                    Treatment  Works
                                                            700,000.000^
                                                            600,000.000
                                                            500,000,000-^
                                                            400,000.000-
                                                            300.000.000^
                                                            200,000.000
                                                            100.000,000
                                                                    .•o

-------
                                            Graph VII.B
                             Percentage of Release* Determined By Each Estimation Method
         Fugitive  Air  Releases
       1987    1988    1989    1990
          Stack Air Releases
                                                                                      Calculations



                                                                                      Measurements
          Releases  to Water
100%^'
 80% -t
 60%
 40%
 20%

       1987    1988    1989    1990
     Releases Injected Underground
100%^
                                                        80%
       1987    1988    1989    1990

-------
        Releases to Land
    1987     1988    1989    1890
Transfers to Publicly Ownec ® N° Baste Reported
                                                                      Works
  100%-
                                                                                     Qther




                                                                                     Emission  Factors



                                                                                     Calculations




                                                                                   O Measurements
Transfers to  Off-Site  Locations
   1987
           1988    1989
                            1990

-------