Date of Award
Master of Science
School of Engineering and Mathematics
Faculty of Communications, Health and Science
Associate Professor James Cross
Time series models have been applied in many areas including economics, stuck recruitment and the environment. Most environmental time series involve highly correlated dependent variables, which makes it difficult to apply conventional regression analysis, Traditionally, regression analysis has been applied to the environmental dependent stock and recruitment relationships for crustacean species in Western Australian fisheries. Alternative models, such as transfer function models and state space models have the potential to provide unproved forecasts for these types of data sets. This dissertation will explore the application of regression models, transfer function models, and state space models to modelling the puerulus stage of the western rock lobster (Panulirus Cynus) in the fisheries of Western Australia. The transfer function models are consulted to examining the influences of the environment on crustacean species and can be used where correlated variables are involved. These models aim at producing short-term forecasts that may help in the management of the fisheries. In comparison with regression models, TFM models gave better forecast values with state space models given the forecast values in the first two years. Overall, it was shown that environmental effects, westerly winds and the Leeuwin Current, have a significant effect on the puerulus settlement for Dongara and Alkimos. It was also shown that westerly winds and spawning stock have a significant effect on the puerulus settlement at the Abrolhos Islands.
Farag, S. A. (1998). A comparison of advanced time series models for environmental dependent stock recruitment of the western rock lobster. Retrieved from https://ro.ecu.edu.au/theses/997