CHAPTER FIVE The nerve center of the ERSP team was the field data management center. Staffed continuously throughout the cleanup by one statistician and one data technician, the data center literally provided overnight data reduction and enabled the resident project manager to give real time advice and technical direction to the cleanup effort. Although the statistical methods were for the most part classical, their application to a massive “brute force” engineering project presented a distinctly non-classical challenge. As decision making rationale and cleanup method evolved, the statisticians regularly visited the field engineering sites to develop an appreciation for the needed format and detail of their advice. Mentioned only briefly at the end of this chapter is the matter of education—but it must be acknowledged as one of the more important contributions of the resident statisticians. The entire ERSP staff and the command and statf of the Task Group as well as members of the DNA commandchain gained their insight into the scientific basis for the cleanup from the data management staff. The technical integrity of the process was largely in their hands. Project Manager’s Note STATISTICS AND DATA HANDLING by Madaline Barnes and Jody Giacomini Desert Research Institute 5.1 INTRODUCTION The Desert Research Institute (DRI), under contract with the Department of Energy, was assigned the responsibility for statistical design and analysis in the Enewetak Cleanup Project, as well as for related data management functions. Because timely information and rapid turnaround on data analyses were critical for keeping the project on schedule, the statistical effort was concentratedin the data processing office on Enewetak Atoll From July 1977 through September 1979 (except for two weeks immediately after Typhoon Alice struck the Atoll), a DRI statistician was present on~island. One Navy data processing technician was assigned to assist the statistician. Although some preliminary computer programming was done and data procedures were established before the project began, most decisions about methods and procedures were made onsite, based on the experience gained as the cleanup progressed. The presence of a statistician on-island facilitated the timeliness of these decisions and also meant that existing procedures could be modified as necessary without delays. In order to allow statistical analyses to be performed using the equipment on-island, a number of simplifications were made in the computer programs. One of the functions of DRI in Las Vegas was to use the first set of data collected on Enewetak to check the accuracy of the simplified routines. Other tasks for which DRI - Las Vegas was responsible included maintaining up-to-date information, transferring IMP spectra to magnetic tape for long-term storage, and performing statistical analyses that were too complex for the computer on-sland. 5.2 STATISTICAL METHODS Most of the statistical techniques used for data on various aspects of the cleanup were from classical Statistics. The major exception was the use of the estimation technique, kriging, to perform the initial surface TRU characterizations. The method, which is discussed more fully in Section 5.2.1, was chosen because the assumptions made are reasonablein light of the physical processes at work, and because it had already proven to yield useful results with radiological data. The kriging approach is also useful because it provides an estimate of the standard deviation of the difference between the true, unknown value at a point and the estimated value at that point. This standard deviation can then be used to give an upper bound on the true value at a specified probability level, thus allowing cleanup criteria to explicitly incorporate a set probability level. 137