HDF4

HDF4 is a legacy data model and software library for storing and organizing large scientific datasets in a single self-describing file. It is part of the Hierarchical Data Format family and was the main iteration before HDF5. The format and its libraries were widely used in atmospheric, oceanographic, and remote-sensing data applications, including NASA’s EOS projects, and many older scientific datasets.

The HDF4 data model centers on a hierarchical structure built from groups called Vgroups, and data containers called Vdata and Scientific Datasets (SDS). Vgroups serve as containers that can hold other Vgroups or data objects; Vdata stores tabular records with named fields; SDS stores multi-dimensional arrays. Attributes provide metadata for files, groups, and datasets. HDF4 supports multiple data types and a library of modules for reading and writing data via a C API, with bindings for Fortran, and later Java.

HDF4 was developed in the 1990s and was maintained for many years, with later development slowing as HDF5 gained prominence. It remains in use in legacy datasets and certain scientific communities, and is supported by The HDF Group. The HDF Group continues to provide HDF4 libraries and tools alongside HDF5, but new projects are typically encouraged to adopt HDF5 or other modern formats. Tools such as HDFView allow browsing HDF4 files, and conversion utilities exist to migrate data to newer formats.

---