Laserfiche WebLink
• • iiiiiiiiiiiiiiiiiii <br />Statistical Sampling for Envvonmental Monitoring Studies: OSA70ffice afTechnologv Transfer. 1998 Ch. XI Pg.l <br />XI COMPOSITE SAMPLING FOR DETECTION OF HOT SPOTS <br />Written By: <br />John W. Kern <br />Introduction <br />An important aspect of reclamation is monitoring of regraded spoils material. The objective of <br />this monitoring is to insure that regraded spoils material is suitable to promote revegetation <br />Typically, a systematic grid of sampling locations is sampled for soil chemistry parameters and if <br />individual samples are above regulatory thresholds, then action is taken, such as removal or <br />burying material thought to be out of compliance. In general, this type of monitoring is based on <br />some sort of systematic sampling grid at relatively close spacings requiring a great deal of <br />laboratory expense to generate the data. In this chapter a compositing technique is described <br />which can cut laboratory assay costs by SO% to 75% while still insuring that all samples which are <br />out of compliance aze located. <br />Inmost compositing schemes, several samples aze mixed to allow better precision for estimation <br />of the mean of the variable of interest. One of the primary drawbacks to this type of compositing <br />is that information on the extremes is lost. In monitoring regraded spoils material, it is precisely <br />the extremes which aze of interest. The basic idea of behind compositing for identification of <br />extremes is actually very simple and was first published by Gore and Patil (1994). <br />Statistical Method <br />Let X„ X„ ... 7{n be a sample of measurements on a variable of interest such as sodium absorption <br />ratio (SAR) in the top foot of regraded spoils at a reclaimed mine site. We can group the n <br />observations into subsets of say k measurements which aze to be combined. An example <br />compositing pattern is shown in Figure 1 where sets of k=4 spatially contiguous samples aze <br />composited in the laboratory. Note that all composites need not contain the same number of sub <br />samples to implement the method, although equal sized composites aze probably easier to keep <br />track of. When the composite samples are assayed one has n/k estimates of mean SAR . The <br />mean in this sense is an average over samples of size 4. To investigate the presence of extreme <br />values the following simple relationship is exploited. <br />