Quality control operates smoothly and almost transparently in those sciences in which experiments are readily reproducible or lead to subsequent experiments that validate the original ones. In the observational sciences, implementing effective quality control for data requires the use of an audit trail system that includes anomaly detection, reporting, and correction, as well as the rigorous refereeing of manuscripts for publication.
Quality assurance, the mechanism used by management to assure that the quality of work is as claimed by those doing it, typically plays a far smaller role in basic science than in applied science and especially in manufacturing. However, one can interpret any mechanism to assure scientific integrity as a kind of quality assurance procedure. This concept would thus include the mechanisms to detect and investigate scientific fraud. Such "quality assurance" efforts are carried out in universities and at the National Institutes of Health, for example.
In recent years, some organizations, such as the Carbon Dioxide Information Analysis Center (CDIAC) at Oak Ridge National Laboratory have devoted significant efforts to producing high quality global Earth science data sets whose accuracy and reliability have been determined, accompanied by the descriptive (metadata) documentation needed for their use. The CDIAC has quality-assured and documented several key global change databases on such diverse topics as concentrations of carbon dioxide and other greenhouse gases in the atmosphere, carbon fluxes from the terrestrial biosphere to the atmosphere resulting from changes in land use, carbon chemistry in the oceans, and long-term climate trends in the United States.32 These value-added data sets are certified as valid by the primary users who collected the data, or by those who subsequently carried out the quality-control checks of the data. This is a somewhat costly, but successful, approach for assuring secondary users of the quality of relevant data sets.33
The trend toward bigger, more complex, and more expensive facilities and programs in the observational sciences, and toward attendant international collaboration, has brought about greater attention to and incentives for effectively archiving data. It also has encouraged the development and maintenance of a curatorial infrastructure necessary to manage the data better, to provide more uniform processing and documentation, and to make retrospective data more easily accessible and usable.34
Research using archived data has grown in scope and importance, especially in enabling the comparison of observations taken at different wavelengths and at different sites. In the space sciences, there are efforts to coordinate data catalogs and indices, facilitating discovery of what data are available (e.g., NASA's Astrophysics Data System and its extragalactic database, SIMBAD). Space as-