The Astronomical Data Warehouse

Abstract

The idea of the astronomical data warehouse has arisen as the natural extension of current trends in astronomical data archives, and from an analysis of the types of query likely to be sent to a virtual observatory. Data warehouses will be centres providing both data and computational facilities. 1 Querying the Virtual Observatory Members of the AstroGrid project have some background in the provision of data archive services and of using them, but in designing the virtual observatory we have been driven principally by some recent collections of typical astronomical problems and use-cases: • The set of some 47 science problems set out on the Astrogrid Wiki pages[1]. • The collection of (so far) 29 use-cases assembled by the US-NVO Project[2]. • A set of 35 queries for the Sloan Digital Sky Survey[3]. Most of these involve a series of interactions with the virtual observatory, which can be further broken down into more basic data access operations. These can be roughly classified as follows: firstly positional queries, second observational queries, and thirdly all other types (which will be called ‘non-positional’ for convenience). 1.1 Positional Queries Positional queries are ones seeking information about a specific small patch of sky, or on a named object. An object name can usually be translated to a position using a name resolver such as Simbad or NED. Given the errors in coordinates, in either case the query is actually one about a small area of sky, typically a circular region about a given (RA, DEC): this has become known as the cone search. Typical examples might be: • Are there any X-ray sources near HD123456? • Can I have all infra-red images centred on PSR0123-456? Because of their importance in astronomy, positional queries make up a high proportion of all queries to on-line resources, but there may be some bias caused