FAIR - Interoperable
Well formatted data that uses discipline conventions and vocabularies, for both the data itself and the metadata used to describe it
Using common non-proprietary data formats
Use common data formats for both metadata and data, avoid using proprietary formats as they are not available to everyone. What is common depends also on your specific scientific community, in climate a widely used formats is NetCDF. A lot of software, standards and conventions used in climate are designed around the NetCDF format.
Follow conventions and standards adopted by your community wherever possible. As climate science includes different disciplines as atmospheric science and oceanography for example, the standards you might have available might change depending on the characteristic of your study.
Controlled vocabularies hold definitions for discipline related terms and concepts. They help ensuring that when data is shared its meaning is clear and explicit. In controlled vocabularies each term is defined by an identifier which also make the vocabularies themselved to be machine-readable.
Reference quality data, metadata, code
A good metadata record will have references to other related records: published data, code, article, grants, projects, institution, researcher identifiers, etc. It is important not only to create links among different records but to clarify the relationships between them as much as possible. There are controlled vocabularies to define the possible relationship which are shared by different data services.