Comparison of short-term rainfall prediction models for real-time flood forecasting

This study compares the accuracy of the: short-term rainfall forecasts obta ined with time-series analysis techniques, using past rainfall depths as th e only input information. The techniques proposed here are linear stochasti c auto-regressive moving-average (ARMA) models, artificial neural networks (ANN) and the non-parametric nearest-neighbours method. The rainfall foreca sts obtained using the considered methods are then routed through a lumped, conceptual, rainfall-runoff model, thus implementing a coupled rainfall-ru noff forecasting procedure for a case study on the Apennines mountains, Ita ly. The study analyses and compares the relative advantages and limitations of each time-series analysis technique, used for issuing rainfall forecast s for lead-times varying from 1 to 6 h. The results also indicate how the c onsidered time-series analysis techniques, and especially those based on th e use of ANN. provide a significant improvement in the flood forecasting ac curacy in comparison to the use of simple rainfall prediction approaches of heuristic type. which are often applied in hydrological practice. (C) 2000 Elsevier Science B.V. All rights reserved.