Wavelet regression models for predicting flood stages in rivers: a case study in Eastern India

摘要: Combining discrete wavelet transform (DWT) and autoregression (AR), two types of regression (WR) models were developed for forecasting 1-day-ahead river stages. In the first type WR models, AR was applied on DWT-obtained subtime series while in second type, modified time which formed by recombining effective ignoring ‘noise’ series. Depending upon different input combinations, five each developed. The efficiency tested monsoon stages Kosi River Bihar State India. During (June to Oct), carries large flow makes entire North unsafe habitation or cultivation. When compared, predicted with greater accuracy than artificial neural network (ANN) purpose. Between gave slightly better results type. best performing model, previous days’ as inputs, River, highest 97.41%, minimum root mean square error 7.9 cm maximum coefficient correlation 0.952.