The Virtual Great Observatory: Do We Know How to Make It Real?

The goal of the SSSC Great Observatory (GO), to use "simultaneous measurements in multiple locations ... to resolve temporal and spatial changes and to understand the interactions of complex systems of regimes", demands a future data environment where users can see data resources as an integrated set. Where data from solar and space physics missions, models and services may be highly distributed, such an integrated view is the paradigm of a Virtual Observatory (VO) or inter-connected Virtual discipline Observatories (VxOs). A first concern to achieving an effective VO/VxO environment must always be that the maximum of potentially useful data and models are openly accessible and made readily findable, with appropriate keywords for search engines and registries of products and services for rapid and uniform access to data from geographically distributed sources in a range of formats and described by different sets of terms. But beyond this, there is no agreed and prioritized list of issues and most critical user requirements for the new data environment. The level of community commitment required to develop consensus on the most urgent needs and to make investigation data usefully accessible to the larger environment will be much deeper than has been seen in the past. At the highest level, we believe that science users want easy-to-use interfaces with reasonable functionality across different types of data and robust performance over a wide range of high resolution, high quality and current data. The skills of both solar/space physics investigators and technologists will be required to build and sustain effective coordination and overall data content. There has been some recent success in providing access to files and some basic services from many distributed sources (e.g., VSO, VSPO, and EGSO). But a possible next level of integration, in which variables are labeled and can be subsetted and supplied in standard ways for use by universal tools and in higher-order queries, will be more difficult and remains today the province of services such as CDAWeb that operate through the backend ingest of data to a common server. Initial experience with the SPASE Data Model is showing that the uniform adoption of terms is likely to be imperfect, and that it is difficult to arrive at clear common goals for the overall "data system." Long-term evolution to a working VO environment will require: (1) a flexible approach to the working architecture where not all elements may make technical sense to distribute; and (2) an effort persisting for some time to give people a chance to try out various partial solutions. An extended effort will also allow time for the large number of products available to be described and made available in uniform ways, although this process is often tedious and not of high priority to data providers. It remains to be seen the extent to which generic tools can significantly supplement individual researchers tool sets, and how robust a system based on many independently maintained interacting parts will be. We will need to encourage long-term financial commitments from supporting agencies and for work across agency boundaries to get past sometimes-awkward initial steps. While the data environment may well be efficient, the results will not be free, and we must develop paradigms for support for efforts that are neither initial research efforts nor developed infrastructure.