Bi-variate Wavelet Autoregressive Model for Multi-step-ahead Forecasting of Fish Catches

Abstract

This paper proposes a hybrid multi-step-ahead forecasting model based on two stages to improve monthly pelagic fish-catch time-series modeling. In the first stage, the stationary wavelet transform is used to separate the raw time series into a high frequency (HF) component and a low frequency (LF) component, whereas the periodicities of each time series is obtained by using the Fourier power spectrum. In the second stage, both the HF and LF components are the inputs into a bi-variate autoregressive model to predict the original time series. We demonstrate the utility of the proposed forecasting model on monthly sardines catches time-series of the coastal zone of Chile for periods from January 1949 to December 2011. Empirical results obtained for 12-month ahead forecasting showed the effectiveness of the proposed hybrid forecasting strategy.