If you want to make earth science data accessible to others in a form that's portable, self-describing, supported by freely available software libraries in multiple languages, and for which community conventions are being maintained, netCDF may be a good choice.

Other alternatives worth considering include

• HDF5 (Hierarchical Data Format 5) is oriented towards high performance computing. It's data model and APIs are more complex than netCDF, but also more powerful.
• Relational database management systems such as Oracle, MySQL, Microsoft SQL Server, or PostgreSQL. Observational data (sensor data, discrete sampling data) often fits the table-based relational model.
• Formats such as Shapefiles and KML/KMZ, appropriate for Geographic Information Systems (GIS), are useful for storing and manipulating spatial data, especially two-dimensional data.
• GRIB (GRIdded Binary) is a WMO standard format for gridded information that is especially good for compression and international data exchange. GRIB data stored on a Unidata THREDDS Data Server can be accessed through a netCDF API.

2010 Unidata NetCDF Workshop > Overview of netCDF