Shih, George B.; Hawkins, Richard H.; Chambers, Melvin D. 1973. Computer modeling of a coniferous forest watershed. Seattle: University of Washington; Coniferous For. Biome Internal Rep. 52. 13 p.
A computer model of watershed behavior Is a detailed accounting of the inputs, storages, transfer functions, and outputs on a drainage basin
basis. The internal processes are modeled as best known within limits of realism and judged importance. As computer programs, they are large and ornate with many branches, checks, and special functions, as are
the natural systems they attempt to mimic.
A characteristic of the system and program is a collection of descriptive parameters or coefficients used in modeling the individual processes,functions, storages, and so on. The more detailed the model, the more parameters must be used. A large part of the modeling process, then,is the selection of proper parameter values to match the observed watershed performance with model performance. The generous supply of parameters in a hydrologic model is both a blessing and a curse. On one hand, it allows more opportunity for a genuine description of the operation of the system. On the other hand, in a statistical sense, it consumes degrees of freedom and invites criticism on the grounds of brute force "curve fitting" and the lack of a unique solution.
To be sure, there is an element of curve fitting involved. However, certain parameters can be either measured or surmised, at least partially. Additionally, knowledge of system operation permits placing limits and realistic ranges on other parameters. Also, human judgment and experienceare used in parameter selection, thus reducing the danger of blind choices even more. Nonetheless, the process is still something less than pure hard science, and contains elements of art and subjective judgment. Crawford (1971) describes hydrologic modeling as a "gray box," or some-thing between a "white box" (pure deductive science) and a "black box" (pure empirical curve fitting).
Despite these shortcomings, computer-based hydrologic models have been used successfully in a wide variety of operational and research situations. Included are the effects of wildfire on streamflow (Fleming 1971), the effects of channelization on flood peaks (Crawford 1969), re-creation ofmissing streamflow records, the effects of weather modification on streamflow (Sopper et al. 1969), and online flood control and reservoir management. The attraction is twofold: (1) economy, and (2) the necessity for otherwise unattainable data estimates, or both.