MONTHLY RUNOFF GENERATION BY NONLINEAR MODELS

A non-linear, physically realizable, discrete runoff model is developed on the basis of the descriptive knowledge of catchment operation, and applied for monthly runoff generation. Linear functional and quadratic functional series are also applied for comparison. A linear programming procedure is introduced into the calibration to ensure the physical realizability of the fitted model. The tests indicate that the linear model is not sufficiently adaptable while the second-order functional series is over-adaptable and consequently unstable, and the best results are produced by the proposed model.