~ - \. ms A -] 1 fo DOCUMENT DUES NUL CONTAIN ECI Reviewed by252 SIVA Date ZZ S9 2 wR ata 8 KG Battelle Pacific Northwest Laboratories ee 409806 September 22, 1976 REPOSITORY U.S. Energy Research and Development Administration Washington, D.C. 20545 . Battelle Boulevard Richland, Washington 99352 Telephone (509) 946-2104 Telex 32-6345 Mr. Tom McCraw Dear Tom: OE PNNA . COLLECTION Marsha |} BOX No. FOLOER Tslands 9685 Ene wre ta ke l¢7G This letter is in response to your request for guidance on the number of samples required for the proposed clean-up survey on the Enewetak atoll. .l_begin with some general comments then discuss specifically the questions you distributed at the meeting in Joe Deal's office on July 29, 1976. There is also an appendix to illustrate the computation of certain confi- dence limits using 239-240py soil data from the island of Janet. This letter has benefited from comments and suggestions by other statisticians at BNW (Drs. Lee Eberhardt, Tony Olsen, and Pam Doctor). The number of samples will depend in part on how well the portable Ge(Li) counter performs in the field, i.e. on how accurately the Ge(Li) readings relate to the amount of plutonium in soil. It will also depend on the statistical design used in the field and on whether it is decided that a contour map of plutonium concentrations is a major goal or whether probability statements about mean concentrations are preferred. Contouring calls for a systematic (uniformly spaced) sampling scheme, while probability statements require random sampling within sub-~areas of an island. Probably it would be desirable to use some kind of sequential sampling scheme, in which results of an initial set of samples are used to decide whether a given area should be (a) considered "clean" (below some standard level), (b) cleaned up, or (c) whether additional samples should be taken before a decision is made. Such a scheme is likely to require continued attention by someone with statistical training, but may be expected to reduce the amount of sampling required. If contouring is used, Dr. Delfiner of the Centre de Morphologie Mathematique, Fontainebleau, France should be consulted on this question of the number of samples required. Dr. Delfiner is knowledgable on "kriging" (a contouring method), and he may be helping Bruce Church set up the technique for use on the islands. We understand that arrangements are being made for Dr. Delfiner to be in Las Vegas for 3 weeks in October and again in November to install his kriging routine on REECo‘s computer. The question of whether In-Situ measurements, soil samples, or both should be used for deciding whether an area or island should be cleanedup requires further discussion. To answer this question we need to know