Some methods of analysis of fishery catch-effort data
Date
1993
Authors
Simons, Clement Mensah
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Most existing methods of analyzing fishery catch-effort data either implicitly assume that the main source of randomness is in the stock dynamics while ignoring randomness in the catching process or vice versa. In this thesis we consider the problem of estimating the parameters of a simple fishery model from a time series of catch and effort by two methods, one of which allows for randomness in both processes.
The Kalman filter approach incorporates a stochastic dynamic model with a stochastic catch model where the only unobservable variable is the stock size. We considered two models in this instance; one in which we assumed that there is randomness in the relationship between the escapement and the stock size and one in which there is none. The main idea is the use of the Kalman filter to obtain the likelihood function in order to estimate model parameters.
Unlike the Kalman filter approach, the contagious distribution method uses a stochastic catch model coupled with a deterministic stock dynamic model. The basic idea followed in this case is the assumption that the probability that a given unit of fish is captured in a given time interval is influenced by other units also being captured.
We applied both methods on real data sets . Data sets were divided into two parts. One was used to fit the model and the other used to test the one-step ahead prediction. In both methods model parameter estimates are
numerically obtained by the method of Maximum Likelihood, and where possible we use the likelihood ratio test procedure to obtain approximate confidence intervals for the parameters. We also estimated the 'optimal effort' using the Kalman filter approach and maximum substainable yield (MSY) using the contagious distribution method.
In terms of the prediction, the methods did reasonably well. Residual plots did not show any sign of model inadequacy. However, for the Kalman filter approach , sometimes we were not able to obtain interval estimates for some of the parameters. Also we were not able to estimate the ratio of the variances and needed to assume it. The method also only estimates stock sizes as fraction of 'equilibrium' stock size.
Description
Keywords
UN SDG 14: Life Below Water