ABSTRACT
Numerous physically based models need a detailed explanation of the overland flow plane at the grid-scale of the numerical model to be used. A detailed description is rarely available, primarily due to time and budgetary constraints. In this chapter, four regressions and two data-driven modelling approaches are applied to model the rainfall-runoff of a mountainous catchment near Palampur in Himachal Pradesh, India. The models are compared based on global goodness-of-fit statistics as well as a visual examination of the time series. The results from these models were compared with polynomial regressions, non-linear regression, fuzzy logic approach, and artificial neural networks approach. The highest value of the coefficient of correlation is obtained for the quadratic regression and thereafter for the fuzzy logic. The results conclude that for the given rainfall-runoff data of a watershed near Palampur, the best correlation method of the rainfall-runoff correlation is dependent upon the obtained value of the coefficient of correlation. In our chapter, the best results for both coefficients of correlation and sum square errors are obtained for the quadratic regression (i.e., polynomial regression of second order). The value of the coefficient of correlation for the fuzzy logic approach is also nearly equal to the quadratic regression method.
