tronomy in the United States has been at the forefront of archiving and distributing data electronically. Archives are an integral part of all U.S. space astronomy missions and are now typically being planned for ground-based observatories and for other countries' space missions as well. The archiving technologies are openly available and shared, and data sets from new and different observations are incorporated increasingly in existing archives. Comprehensive catalogs and good user-access tools are recognized as very important, as are properly maintained and preserved duplicate data sets. Some major archives are also duplicated at different sites to reduce communication loads and to promote innovation and allow different uses.
In the Earth sciences, the study of Earth processes involves time-dependent behavior over time scales ranging from seconds to millions of years. For relatively short time scales (years or less), observational data from a common observing platform may be available from a single database. For longer time scales (decades to centuries to many thousands of years), it is necessary to scrutinize all of the retrospective observations available and to use proxy data preserved in the geologic record or in written records. Research on global change and on natural hazards, for example, whose goal is improved prediction of future conditions or events, depends heavily on accurately reconstructing the record of the past. Box 3.2 provides several examples of interesting data reconstruction projects in China.
Because there has been an increasing awareness of the great value of retrospective Earth science data, some conscientious efforts to rescue and preserve older data are being made both nationally and internationally. In the United States for example, the National Climatic Data Center and the National Geophysical Data Center have devoted time and resources to data rescue in recent years and now have a policy of transferring all their digital data holdings to new storage media at least once every 10 years.
Many specialized observational databases requiring long-term retention exist in biology as well. These include such diverse subjects as agricultural records of many types, including experimental field tests going back to the last century; museum, zoo, herbaria, and microbial culture collection records; hospital and other medical records; ecological data; breeding histories of domestic animals and plants; macromolecular sequences and their accompanying annotations; taxonomic treatments, toxicological information; folk medicine; and characterization of biological products such as food, fiber, and fine and bulk chemicals. Some of these data are in computers, but require normalization for consistency and readability. Others could be transformed into machine-readable form to make them generally accessible. These two tasks are labor intensive and require a large component of highly skilled labor. Such undertakings require careful prior evaluation for potential worth. Selective evaluation and support is necessary not only to enhance the intellectual effort to maintain existing databases, but also enable the creation of new ones.
Once the primary data are analyzed and used to publish research results, the