1. An interface to build and launch experimental forecast workflows that run WRF and other forecast and data analysis codes on remote resources. A drop-and-drag graphical composition tool can be used to build new workflows and data analysis tools and to visualize the execution progress of the computation.

2. A private metadata catalog of all the experiments and data products that they have used or generated. The metadata contains searchable descriptions of each object and its provenance as will as links to the actual data.

3. A tool to search available weather data based on geographic region, data attributes and time.

4. A ontology-based Glossary of weather-related terms and links to educational modules to guide student learning.

The portal is actually a front-end “Gateway” to the TeraGrid and Unidata weather data resources. TeraGrid is NSF's network of supercomputers that provide the computational and high-end data storage resources for the nation's largest computational science projects. Both TeraGrid and the LEAD portal are based on a service-oriented architecture. Every aspect of the portal's operation is based on interaction with a distributed network of web services. These includes services that crawl the Unidata resources indexing its content, services that manage data storage and the metadata catalog and event notification services that allow all the services to communicate and generate a providence history of each of the users experiments. Scientific applications like WRF are also managed as services and the workflow system orchestrates these application and data movement services using a web service based workflow system called BPEL. This paper will provide a complete overview of the portal and an in-depth look at the way workflows are constructed and executed using the service architecture.

If the Internet is available from the conference room, the talk will feature a live demo of the portal, data subsystem and workflow enactment tools.