Notas Científicas Comparação de modelos matemáticos não-lineares empregados na análise de desempenho de sistemas de irrigação

Various non-linear mathematical models utilized in the analysis of irrigation system performance were compared aiming to indicate the model which better represents applied water distribution profiles. Four probability models (Normal, Log-normal, Gamma and Beta) as well as two potential models (Silva's and Karmeli's models) were considered, and were applied to 91 cases of irrigation performance evaluation. The pairwise comparison of the cumulative frequency distribution curves of the sum of square errors, obtained from fitting each model to the observed data, revealed that Silva's model is the best among the non-linear models tested.