Sequential geostatistical simulation methods to assess the error distribution function for biomass determined from acoustic survey data

No abstracts are to be cited without prior reference to the author.

Sequential geostatistical simulation methods can be used to calculate empirical confidence intervals for biomass determined from acoustic surveys. Such simulation methods have several advantages: precision can be estimated in the presence of autocorrelation and non-random sampling and the combined variance of the acoustic and fish length-frequency measurements can be estimated empirically rather than analytically. In addition, an empirical probability density function (pdf) of the abundance estimate is generated, and can be used, for example, to determine the precision of indices that characterize spatial distribution, such as center of gravity. Examples from the Eastern Bering Sea and the Gulf of Alaska will be presented. The method can be extended to include other sources of uncertainty so that the total error is more closely approximated. A second method for evaluating total uncertainty, consisting of virtual sampling from a known fish distribution (created by simulation) is also being explored.