Multistep-ahead flood forecasting using wavelet and data-driven methods

Accurate forecasting of floods is vital for developing a flood warning systems, flood prevention, flood damage mitigation, soil erosion reduction and soil conservation. The objective of this study is to apply two hybrid models for flood forecasting and investigate their accuracy for different lead times. These two models are the Wavelet-based Artificial Neural Network (WANN) and the Wavelet-based Adaptive Neuro-Fuzzy Inference System (WANFIS). Wavelet decomposition is employed to decompose the flood time series into approximation and detail components. These decomposed time series are then used as inputs of Artificial Neural Network (ANN) and adaptive Neuro-Fuzzy Inference System (ANFIS) modules in the WANN and WANFIS models, respectively. The WANN and WANFIS models yielded better results than the ANN and ANFIS models for different lead times. The WANN and WANFIS models performed almost similarly. However, in terms of model efficiency, the WANFIS model was superior to other models for lead times of 1 to 6 hours, and the WANN model was superior to other models for lead time of 8 to 10 hours. The results obtained from this study indicate that the combination of wavelet decomposition and data-driven models, including ANN and ANFIS, can improve the efficiency of data-driven models. Results also indicate that the combination of wavelet decomposition and data-driven models can be a potential tool for forecasting flood stage more accurately.