Comment on “Physically based hydrologic modeling: 2, Is the concept realistic?” by R. B. Grayson, I. D. Moore, and T. A. McMahon

In the first of two companion papers, Grayson et al. [1992a] described the application of a relatively detailed "terrain-based" model to two quite different catchments. One is in Australia, characterized by both surface and subsurface runoff generation mechanisms, and the other is a catchment in southern Arizona with exclusively surface runoff processes. The second paper [Grayson et al., 1992b], which is the subject of this comment, is an extensive and critical discussion of results of the first paper. It includes a rather pessimistic assessment of the results of this model exercise, but consists primarily of opinions concerning the use of "physically based" models "intended for use as hydrologic components of sediment and nutrient transport models" (p. 2659). Hereafter we refer to these two papers as paper 1 and paper 2. It is our view that the authors' assessment of their results is overly pessimistic, and many of their rather sweeping generalizations therefrom are unwarranted. While some good points were made by the authors (they have mostly been made before), it is hard not to conclude after reading their philosophizing that the authors came to their exercise of detailed "physically based" modeling with unrealistic expectations. Much similar philosophy has appeared recently on physi cally based model applicability [Beven, 1989; Dunne, 1982; Klemes, 1988;Loague and Freeze, 1985], and most of that is referenced in the subject manuscript. The work by Loague and Freeze [1985] is a classic example of apples versus oranges, in which calibrated empirical models were indi cated to be apparently better than uncalibrated physically based models, and has often been cited to cast doubt on physically based models. In reading all the philosophical criticisms of the "failures" and limitations of "physically based" models, it seems clear that some have indeed ex pected such models to be exact mimics of nature. But one also wonders whether such expectations have resided more in model developers or in model users, and one wonders how many model users have shared such unrealistic expecta