This paper presents the design and the architecture of the `Solar Web Project' which aims to provide scientists interested in solar data with collaborative software for browsing distributed and heterogeneous solar databases using a unique web-based interface.
The project is developed at MEDOC (Multi-Experiment Data and Operations Centre), located at the Institut d'Astrophysique Spatiale (Orsay, France) which is one of the European Archives for SOHO (selected by ESA) and for TRACE data. JAVA and XML are technologies being used for designing and developing this tool.
Today scientists interested in solar science are facing large amounts of data and many heterogeneous tools to identify and analyze them. Observations often arise from collaborative programs involving several observatories (space and/or ground-based). Cross-data analysis starts then with a heavy task: to identify and to retrieve data around several archive centers, each of them using different user interfaces, vocabularies, and semantics.
The tool developed for the SOHO archive (Scholl 1998) at MEDOC (Scholl 1999) is a first step in the process of simplifying the access to heterogeneous data. SOHO is a spacecraft that has on-board 12 instruments, each of them having their own catalog and data in their own format. The software developed to browse the SOHO catalog and to retrieve files (including a web-based graphical user interface) was designed with open concepts in mind: this tool had to be generic enough to be able to easily manage 1) homogeneously all instrument catalogs and 2) other solar observations than those from SOHO (TRACE data are already available with this tool).
The Solar Web Project (Scholl 2001) ensues from: the acknowlegment that scientists have no unique or generic tool available for browsing several heterogeneous and worldwide distributed archives, the fact that an archive center should not store all mission data (data should be located near the relevant scientific experts), and the need to extend the SOHO archive access system to other solar archives in order to help scientists in their data search phase.
The aim of the `Solar Web' tool is to provide users with one unique interface to access several heterogeneous and distributed archives. This assumes the development of a software with capabilities of sending multiple queries over the Internet, waiting for remote results before merging them and finally displaying them on one single view. But it also assumes that these archives are coordinated in order to standardize communications and catalogs.
The project is divided in 2 main parts: 1) scientific work: definition of a common model based on a standardized vocabulary for describing a `Generic Solar Observation', and 2) technical work: software design and development.
The first step is to defined what a `Generic Solar Observation' means. This phase consists in classifying a few parameters which are sufficient for the purpose of selecting the data. Around twenty keywords are grouped in 4 categories: General Parameters, Observational Technical Parameters, Scientific Parameters, and Data Management Parameters.
The purpose of this model is to help communications between existing archives. By using such a model, existing archives will be able to communicate in a standard way, even if some of them would probably not have all this information or will have to convert their own keywords to this model.
This software is based on a 3-tier and distributed architecture. It has been designed to be open to new kinds of data (see Figure 2).
On the server side, the `Application Server' fulfils the following roles:
On the client side, the applet is dedicated to the powerful query capture and result visualization. The client (excepted for the dialog with the server) and the query generator are the only business specific components. These two modules will be available as an API of the generic software to implement extensions.
Exchanges between the main server and remote applications are made using either a native application protocol to simplify development, or XML with a standardization objective: the `Generic Solar Observation' is implemented as an XML DTD.
Security is a major concern in this project. Permitting direct remote accesses is always a potential security risk for any organization. A secure protocol for communications between remote servers is currently under study.
`Solar Web' is currently a stand alone development effort from MEDOC. Such another project exists, the Virtual Solar Observatory (Hill 2001), but it is still at a very early stage. Discussions are in progress with potential partners. Tests will be made with Bass2000 (Themis data, at Tarbes, France) to experiment with the communication between MEDOC and an existing remote web-server. The status of the project can be found at http://www.medoc-ias.u-psud.fr/ archive/solarweb/ . We are open to suggestions for new collaborations.
Scholl, I., et al. 1998, in ASP Conf. Ser., Vol. 172, Astronomical Data Analysis Software and Systems VIII, ed. David M. Mehringer, Raymond L. Plante, & Douglas A. Roberts (San Francisco: ASP), 253
Scholl, I. 1999, in Proceedings 8th SOHO Workshop, (ESA Publ. SP-446) (Paris 1999), 611
Hill, F. 2001, in ASP Conf. Ser., Vol. 225, Virtual Observatories of the Future, ed. Robert J. Brunner, S. George Djorgovski, & Alexander Szalay (San Francisco: ASP), in press
Scholl, I. 2001, in ASP Conf. Ser., Vol. 225, Virtual Observatories of the Future, ed. Robert J. Brunner, S. George Djorgovski, & Alexander Szalay (San Francisco: ASP), in press