Journals Proceedings

Abstract

Ongoing needs to achieve the best accuracy of flood forecasting has stimulate this study to investigates the potential of two data driven model, where their application are relatively new in hydrology problems. The approaches studied here are KNearest Neighbours (KNN) and Multivariate Adaptive Regression Splines (MARS). To analyze and compares the performance of these two approaches in flood prediction, Pahang River in Malaysia has been selected as area of study. 30-years historical data set of daily rainfall and runoff at upstream tributaries of Pahang River were used to develop and validate the capability of both approaches in one-year-ahead prediction of flood discharge. The effect of different length of record data to the performance of models was also examined. Simulation results showed that longer period data can provide significant improvement to the performance of both approaches. However, satisfactory result of flood prediction only appeared superior for MARS model.