GROUND WATER--Paper entitled, "Ground-water models cannot be validated," by Leonard F. Konikow and John D. Bredehoeft.


In Reply Refer To:                       February 5, 1993
Mail Stop 411


OFFICE OF GROUND WATER TECHNICAL MEMORANDUM NO. 93.04

Subject:  GROUND WATER--Paper entitled, "Ground-water models
          cannot be validated," by Leonard F. Konikow and
          John D. Bredehoeft.

Ground-water models of various types and complexities have become
accepted and widely used in the Water Resources Division over the
past several decades.  As we use these important tools and report
on their results, it is essential that a proper perspective on
their validity and uncertainty is maintained.

The subject paper, published in Advances in Water Resources,
volume 15, no. 1, 1992, p. 75-83, addresses a number of concerns
with respect to model errors, model uncertainty, and model
testing.  The paper also takes issue with use of the terms
validation and verification.  The last paragraph of the paper
summarizes well their argument:

     What is usually done in testing the predictive capability of
     a model is best characterized as calibration or history
     matching; it is only a limited demonstration of the
     reliability of the model.  We believe the terms validation
     and verification have little or no place in ground-water
     science; these terms lead to a false impression of model
     capability.  More meaningful descriptors of the process
     include model testing, model evaluation, model calibration,
     sensitivity testing, benchmarking, history matching, and
     parameter estimation.  Use of these terms will help to shift
     emphasis towards understanding complex hydrogeological
     systems and away from building false confidence into model
     predictions.

We encourage all users of ground-water models to read the paper by
Konikow and Bredehoeft carefully.  Furthermore, the terminology
they suggest should be adopted in reporting on ground-water model
results.  The use of one term instead of another will not affect
the level of confidence that is justified for a particular
application of a model of a physical system, but the terminology
used might be important in adequately conveying to the reader and
reminding ourselves of the level of confidence in the model that
is appropriate.



                             Acting Chief, Office
                             of Ground Water

Attachments

Distribution:  A and B
               S (NR (20), CR (35), WR (5)), FO and PO
               (memo and report)