Title

Propagation of measurement uncertainty in spatial characterisation of recreational fishing catch rates using logistic transform indicator kriging

Document Type

Conference Proceeding

Publisher

Modelling and Simulation Society of Australia and New Zealand

School

School of Engineering / Centre for Marine Ecosystem Research, School of Natural Sciences

RAS ID

20194

Comments

Originally published as: Aidoo, E.N., Mueller, U.A., Hyndes, G.A., Ryan, K. (2015). Propagation of measurement uncertainty in spatial characterisation of recreational fishing catch rates using logistic transform indicator kriging. In the proceedings of MODSIM2015, 21st International Congress on Modelling and Simulation. (pp. 145-151). Gold Coast, Australia. Original paper available here

Abstract

Geostatistical estimation techniques (such as kriging) have been widely accepted and applied to characterise the spatial distribution of natural phenomena. In fisheries science, these techniques have been applied for computing indices including catch per unit effort (catch rate) used for stock assessment. The nonparametric kriging approach, known as indicator kriging, is particularly helpful for the estimation of catch rates using data observed from recreational fishing surveys because it can also handle other features of the data distribution such as zero-inflation, high skewness and class-specific spatial patterns. The problem considered in this paper is the use of indicator kriging for estimation of catch rates associated with an uncertainty due to multiple measurements observed at some locations. This measurement uncertainty is often non-negligible and needs to be propagated to produce accurate estimates. In addition, the uncertainty might be spatially autocorrelated and correlated with the data and so will restrict the use of a parametric approach of uncertainty propagation. Using catch rate data for Australian herring (Arripis georgianus) from a recreational fishing survey, this study presents a soft indicator kriging approach that uses a logistic function transformation to allow the propagation of measurement uncertainty in the estimation. The performance of the uncertainty propagated model was evaluated based on the leave-one-out cross-validation method. The accuracy plot and goodness statistic indicate agreement between the expected and empirical proportions of the observed catch rates falling within probability intervals of increasing size. These suggest a good estimation performance of the approach for Australian herring catch rate data. The spatial distributions of catch rate estimates with propagated uncertainty can be used for quantifying location-specific patterns to assist stock assessment or validation of policy supporting stock assessment models. The measurement uncertainty is considered to be potentially valuable for estimation of catch rate.

Access Rights

Free to access

Share

 
COinS