SAMPLING STRATEGY TO DETECT A CHANGE IN CONCENTRATION OF TRACE ORGANIC CONTAMINANTS IN MARINE SEDIMENT

The ability to detect changes in the environmental level of contaminan ts depends on the variances associated with the measurements. The sour ces of variance contributing to the determination of chlorobiphenyls a nd organochlorine pesticides in sediment from the Garroch Head sewage sludge disposal site have been examined. The total variance is conside red to be the sum of a 'field variance' associated with this site, and long-term analytical variance. The mean short-term coefficient of var iance (CV) for analysis of seven chlorobiphenyls (CBs) and five pestic ides was 10.4% and 14.4%, respectively. Long-term analytical variance was higher at 21% and 42% for CBs and pesticides. The field variance a t the Garroch Head site was estimated from this data and analytical da ta from five replicate sediment cores by ANOVA. Field variance was 32. 7% for CBs and 28.4% for pesticides. Using these estimates for varianc e components contributing to the overall variance of the measurements, the effect of varying sampling and analytical procedures on measureme nt precision and cost factors has been examined. Field variance genera lly dominates the overall measurement variance, such that improvement of the long-term analytical variance below 20-30% is of little benefit . Single or duplicate analysis and sample pooling are generally the mo st cost effective strategy.