Modelling and forecasting cyclical fish landings: SETARMA nonlinear time-series approach

Several economically important marine fish species like mackerel, oilsardine, and Bombayduck exhibit cyclical fluctuations in landings. Modelling and forecasting of such data sets have generally been carried out using the well-known autoregressive integrated moving average (ARIMA) methodology. The main limitations of this methodology, viz., assumptions of stationarity and linearity are pointed out. The purpose of this paper is to bring to the notice of fishery biologists, the existence of a very versatile self-exciting threshold autoregressive moving average (SETARMA) nonlinear time-series model, which is capable of describing cyclical fluctuations. This model is briefly discussed followed by a description of a powerful stochastic optimization technique of real-coded genetic algorithms (RCGA) for fitting of this model. The relevant computer program is developed in C - language. Finally, as an illustration, modelling mackerel landing data of Karnataka state for the period 1961-2008 is considered.