The design for new observatories, facilities, or instruments need to include an end-to-end Science Data Management Plan (SDMP) that describes how the data will be produced, starting from the proposal stage through observation, how it will be handled and delivered to the science user, and how that user will produce science results from those data. Each step of the plan must be shown to be feasible. A number of issues need to be understood in order to define the scope of the data management that is required. Does the SDMP include an adequate system for collecting and managing metadata at the time of observation ? Does the SDMP require that the observatory be involved in developing data processing software? Does the observatory need to produce processed data products? Does the plan require that the primary science user is able to combine their proprietary observations with other data sources using VO systems? Do the science data have sufficient long-term value that an archive is required? The answers to these questions are unique to each observatory. There is no general set of answers. The requirements for a cosmic microwave background experiment will be very different from a massively-multiplexing spectrograph used to collect observations of stars or galaxies. But in every situation a set of science goals will have been defined and the adopted data management model needs to be support the achievement of those science goals. Observatories that need to support integration of their data into the virtual observatory need to plan for that. They should be aware that significant costs are associated with this need. The Virtual Observatory capabilities do not come for free. The quest for VO compatibility produces pressures on the entire observatory system from the proposal preparation stage, through the observation plan, and on through the data management system. Designing a coherent system is a major challenge.
展开▼