An evaluation of the methodology for prediction of capelin biomass

Abstract

The one-year prediction of capelin made as part of the assessment after the annual joint acoustic capelin survey during the autumn, is evaluated. Such predictions were made for the period from 1981 till present, and compared to observed stock sizes. The results show that the predictions underestimate the stock size in about half of the years and overestimate the stock size in the others, but in 18 out of the 23 years the observed stock sizes are within the 90% confidence interval of the predictions. It is found that there is a tendency for overestimating stock size in periods when the stock decreases and vice versa. The ratio between predicted and observed stock sizes is variable and some times quite high for stock sizes below one million tonnes (collapsed stock size) but varies between about 0.5 and 1.5 and is unrelated to stock size for larger stock sizes. These results are discussed in light of the various components of the models involved in the predictions.