•V
                       UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
                                     WASHINGTON, D.C,  £0460
            'eceister 17, 1984

           Honorable William D. Ruckelshaus
           Administrator
           U.S. Environmental Protection Agency                              OFFICE OF
           .-,.._       ,__               S                               TM6 ftOM4NlSTRAT
           401 M Street, SW
           Washington, B.C.   2Q460

           Dear Mr.  Suckelshaus:

                The Environmental Protection Agencyt  in the implementation of its
           authorizing statutes,  utilizes a number of modeling techniques to develop
           and enforce regulations and  standards for various pollution sources.   SPA
           utilizes models for at least three .major predictive purposes.   These
           include 1) as a supplement to, and sometimes substitute for, direct measure
           ments;  2) to clarify environmental mechanisms and processes; and 3) as a
           tool for the development of  policy guidance.  Examples of the application
           of modeling by the Agency include the projection of exposures from land-
           based incinerators of  liquid hazardous wastes and the determination of
           industrial source compliance with the National Ambient Air Quality
           Standards-

                The widespread use of models by EPA,  the states and the regulated
           community has major economic implications.  More attention should be
           directed to the evaluation of the uncertainity and confidence limits  to
           be applied to model results*   It is  also important to identify the
           relative fraction of the uncertainty that  can be attributed to the various
           subcomponents of large, complex models.  Without explicit identification
           of the  assumptions and limitations of the  submodels, it is impossible for
           one to  judge whether or not  a large  model  is consistent  overall.

                From time to time the Science Advisory Board (SAB)  is asked to
           review  models developed or used by the Agency.  Many of  these models  have
           not been adequately tested for their reliability.   Nor has the proper mix
           between monitoring and modeling been examined systematically.   All models
           adopt simplifying assumptions and are therefore incomplete.  A model
           validation effort should center around defining the extent of  these
           limitations.

                Validation of these pivotal procedures is a priority issue for EPA's
           research agenda.   In view of  this situation, the SAB makes the following
           recommendations;

-------
                                                                                     \
     o that EPA initiate a systematic effort at model validation.  Given
       the Agency's increasing reliance upon exposure assessment in
       making risk assessment and risk management decisions, the SAB
       recommends that the Agency focus its efforts initially on the
       validation of exposure models*
     o
that such an exposure model -validation effort address the appropriate
mix "between monitoring and modeling.
     o that EPA seek to determine the relative utility of exposure    ':
       modeling approaches for several biological media and systems that
       are of regulatory priority*

     o that IPA prepare an analytical rationale for this effort in the
       form of a case study that would discuss the approaches to be taken
       and milestones by which to measure the degree of success.

     o that the Agency seek SAB review of the model validation effort
       during the life of the project.

     A first step toward implementation of this initative would be to
submit for SAB review a specific model, now in use or about to be
introducedj so that its efficacy may be independently evaluated, including
whatever progress EPA itself has nade la validating it and exploring its
applicability to other situations.

     The Board fully recognizes that such an undertaking will require
ample resources.  By not taking a leadership position on the need to
validate models, however, the Agency takes an even larger risk that many
of its current procedures for assessing risk will soon be rendered obsolete,
The SAB is concerned that.this is already happening as current models are
utilized for purposes for which they were not designed and for which
there is insufficient scientific justification.  The continuation and
extension of this practice will cause further strains upon the Agency's
scientific credibility.

     I would like to discuss these SAB observations and recommendations
with you and your successor at a mutally convenient time*

                                    Sincerely,       ,


                                              -      Us-
                                                    / \ V. ^ •'-

                                    Norton Nelson, Chairman
                                    Executive Committee
                                    Science Advisory Board
cc:  Alvin Aim

-------