The THREDDS project has developed a framework for publishing, cataloging, describing, and discovering distributed scientific datasets. The framework concentrates on communicating catalog information between data providers and data users. Existing metadata standards are used for describing datasets. Data access protocols like DODS and ADDE are used for accessing the underlying data.

This paper will detail various pieces of the THREDDS framework: 1) the XML documents used; 2) the Java library that has been developed for creating and interpreting these XML documents; 3) several prototype applications that are using the THREDDS framework; and 4) various discovery services available for THREDDS datasets.