Laboratory performance in the Sediment Laboratory Quality-Assurance Project, 1996-98

This report describes the results of the first 3 years of an ongoing study of sediment laboratories used by the U.S. Geological Survey (USGS). The sediment laboratories currently in operation constitute the entire USGS national laboratory system for sediment analyses. As with all environmental data, physical sediment data can be expected to contain a certain amount of difference, whether the difference resulted from the collecting, transporting, or analyzing of the sample. Because it is important to know how each laboratory is performing, the USGS initiated the Sediment Laboratory Quality-Assurance (SLQA) project in August 1996. From August 1996 through June 1998, five studies were made, the results of which are the basis of this report.

The focus of the project is on quantitative analyses done on water-sediment mixtures to derive suspended-sediment concentrations, sediment-mass determinations, and sand/fine separations. For the purpose of this report, all mass determinations are net values—the tare weight of the container is excluded. The fine-size material is defined as particles sieved to a size of less than 62 micrometers (µm) and sand-size material is defined as particles sieved to a size of between 63 and 125 µm. Also, in this report, class 1 samples are defined as samples containing 50 to 100 milligrams (mg) of fine-size material, class 2 samples are defined as samples containing 101 to 300 mg of fine-size material, and class 3 samples are defined as samples containing 2,200 to 3,200 mg of fine-size material. In studies 96-1 through 98-1, the amount of sand added to each sample ranged from 9 to 28 percent of the mass of fine-size material in each sample class.

Analytical results from all sediment quality-control samples are compiled and statistically summarized by the USGS, Branch of Quality Systems, both on an intra- and interlaboratory basis. When evaluating these data, the reader needs to keep in mind that every measurement has an error component associated with it. It is premature to use the data from the first five SLQA studies to judge any of the laboratories as performing in an unacceptable manner. There were, however, some notable differences in the results for the 12 laboratories that participated in the five SLQA studies. For example, the overall median percent difference for suspended-sediment concentration on an individual laboratory basis ranged from –18.04 to –0.33 percent. Five of the 12 laboratories had an overall median percent difference for suspended-sediment concentration of –2.02 to –0.33 percent. There was less variability in the median difference for the measured fine-size material mass. The overall median percent difference for fine-size material mass ranged from –10.11 to –4.27 percent. Except for one laboratory, the median difference for fine-size material mass was within a fairly narrow range of –6.76 to –4.27 percent. The median percent difference for sand-size material mass differed among laboratories more than any other physical sediment property measured in the study. The overall median percent difference for the sand-size material mass ranged from –1.49 percent to 26.39 percent. Five of the nine laboratories that do sand/fine separations had overall median percent differences that ranged from –1.49 to 2.98 percent for sand-size material mass. Careful review of the data reveals that certain laboratories consistently produced data within statistical control limits for some or all of the physical sediment properties measured in this study, whereas other laboratories occasionally produced data that exceeded the control limits.