The search for a better (or "best") parameter set is facilitated greatly by specifying upper and lower limits for each of the parameters—this defines the "feasible" region in which an "optimal" parameter set is expected to lie. A useful perspective is to consider this feasible region to be the initial uncertainty in the parameter estimates, based on available prior information. For certain parameters, these upper and lower limits are easily selected based on physical considerations related to the characteristics of the watershed. For parameters such as equation exponents defining the degree of nonlinearity of a transformation, one may only be able to guess at an appropriate range of values. The selection of constraints is rather model dependent and can be quite subjective. However, if the parameter adjustment procedure is sufficiently powerful, the impact of this subjectivity should be minimal.
Was this article helpful?