Th-2104B-8
Cokriging As a Method for Combining Resource Surveys, Decreasing Uncertainty, and Mitigating Bias

Thursday, August 21, 2014: 11:10 AM
2104B (Centre des congrès de Québec // Québec City Convention Centre)
Burton Shank , Population Dynamics Branch, NOAA Northeast Fisheries Science Center, Woods Hole, MA
Jui-Han Chang , Population Dynamics Branch, NOAA Northeast Fisheries Science Center, Woods Hole, MA
Dvora Hart , Population Dynamics Branch, NOAA Northeast Fisheries Science Center, Woods Hole, MA
Karen Bolles , Arnies Fisheries, Inc., New Bedford, MA
William DuPaul , Virginia Institute of Marine Science, Gloucester Point, VA
Scott M. Gallager , Biology Department,, Woods Hole Oceanographic Institution, Woods Hole, MA
David Rudders , Virginia Institute of Marine Science, Gloucester Point, VA
Richard Taylor , Arnies Fisheries, Inc., New Bedford, MA
Norman Vine , Arnies Fisheries, Inc., New Bedford, MA
Amber D. York , Biology Department,, Woods Hole Oceanographic Institution, Woods Hole, MA
Fishery scientists may receive resource survey data from multiple sources. However, the different surveys may use different survey gear, not cover the entire resource, may survey different areas in different years, or not span enough years to be included as an independent survey index. In a fishery where the spatial distribution of biomass is reasonably conserved across surveys within a season, cokriging, a geostatistical technique, is a potential method for combining overlapping surveys or using high-resolution, small-scale surveys to improve lower-resolution, large scale surveys. For the sea scallop fishery in the Northeast US, industry-supported research by multiple research groups provides survey data that augment the annual federal survey. However, differences in spatial extent, survey methodology, and sampling gear complicate unifying the survey data into a single biomass estimate. We examine the use of cokriging for using the industry-supported survey data to improve the biomass estimates for the federal resource survey. We find that this method consistently decreases survey CVs, often dramatically, and helps to remove bias for areas where surveys are spatially incomplete. We conclude that co-kriging is a promising technique for better utilizing survey data in stock assessment when such data cannot be used as an independent survey index.