Research Document 2017/021

Evaluating Models to Forecast Return Timing and Diversion Rate of Fraser Sockeye Salmon

By Folkes, M.J.P., Thomson, R.E., and Hourston, R.A.S.

Abstract

Management of the Fraser River sockeye fishery includes a pre-season planning component that relies on the forecast of three variables that represent characteristics of the returning adult run: recruitment, migration timing to local waters, and migration entry route relative to Vancouver Island (as defined by the Northern Diversion Rate). In this paper, we evaluate the two components related to forecasting the homing migration of adult Fraser sockeye. These forecasts are based on statistical relationships between observed and modeled environmental variables and the known migratory patterns that they are assumed to influence. We present the results from several software tools developed by the authors to identify a suite of North Pacific oceanic time series that have biologically relevant relations to both return timing of two Fraser sockeye stocks and northern diversion rate. In addition to the models that are founded on assumed mechanistic connections, we evaluate a series of naïve models based on the statistics of the dependent time series. Separate from the forecast model evaluation, we also explore the potential influence of major El Niño events on stock migratory behaviour. All forecast models are evaluated by performance analyses that appraise the forecast precision, accuracy, and robustness in relation to seasonal to interannual changes in the time series. We introduce a new method of ranking performance metrics that provides a better indication of a model’s relative rank. The final stage of model selection is based on tolerance curves, which are isopleths depicting the number of models that fulfil a desired forecast uncertainty at a given level of likelihood. These tolerance curves serve as an objective tool that will help bridge the science of statistical model development to the subjective requirements of fisheries management.

Complete document (PDF, 19,369 KB)

Accessibility Notice

This document is available in PDF format. If the document is not accessible to you, please contact the Secretariat to obtain another appropriate format, such as regular print, large print, Braille or audio version.