Geomorphology-based artificial neural networks (GANNs) for estimation of direct runoff over watersheds

Focusing on the problem of estimating direct runoff over a watershed resulting from rainfall excess, the goal of this study is to develop an artificial neural network (ANN) that explicitly accounts within its architecture for the geomorphologic characteristics of the watershed. Such a geomorphology-based artificial neural network (GANN) is utilized to estimate runoff hydrographs from several storms over two Indiana watersheds. The architecture of the GANNs as well as a part of the network connection strengths are determined by watershed geomorphology, leading to a parsimonious ANN modeling tool. Comparisons of validation results from the GANN model with observed hydrographs over several events for two watersheds are presented. Results obtained by using the geomorphologic unit hydrograph theory (GIUH) are also included for illustration purposes. This study reveals GANNs to be promising tools for estimating direct runoff. (C) 2003 Elsevier Science B.V. All rights reserved.