A Transfer Function Model with Effort, Catch Rates and Seasonal ARIMA Errors to Predict Monthly Catches for the Esperance Southern Rock Lobster Fishery
by
Monty Craine
Department of Fisheries, WA
Coauthors: Y.W. Cheng (WACEIO, Curtin University), J. Prescott (South Australia Research and Development Institute)

Catches for the southern rock lobster fishery in the Esperance area ( 120 ^o E to 125 ^o E) of Western Australia are difficult to predict. The CPUE series for the fishery exhibits switching behaviour between sequences of high and low values. We show that annual catches are highly correlated with lagged interactions involving catch rate data in May from the central area of the northern zone of the South Australian southern rock lobster fishery and a yearly moving average of the Fremantle Sea Level indicator. The formulation emulates the biological possibility that larvae may be transported from South Australian waters to the Esperance area during a one- to two-year timeframe. Monthly catches are predicted using a seasonal ARIMAX model, where the transfer function is defined by the annual specification. The model is validated by forecasting three years of monthly catches (1997/98 to 1999/2000). The transfer function component is highly significant for the seasonal ARIMAX model. Further techniques to enhance the error forecasts and to face the problem of negative 95% confidence interval limits are discussed.

Date received: November 13, 2001