Application of wavelet transform in runoff sequence analysis*
Suyi Liu, Xianzhang Quan, Yongchuan Zhang · Progress in Natural Science Materials International · 2003
A wavelet transform is applied to runoff analysis to obtain the composition of the runoff sequence and to forecast future runoff. An observed runoff sequence is firstly decomposed and reconstructed by wavelet transform and its expanding tendency is derived. Then, the runoff sequence is forecasted by the back propagation artificial neural networks (BPANN) and by a wavelet transform combined with BPANN. The earlier researches seldom involve the problem of how to choose wavelet function, which is important and cannot be ignored when the wavelet transform is used.With application of the developed approach to the analysis of runoff sequence, several kinds of wavelet functions have been tested.