Geosciences data is the important basis for science research and social development. In the past decades, many countries including developed countries and developing countries paid more attention on geosciences data sharing and related e-infrastructure construction. Because geosciences data have typical multi-sources, multi-types, cross disciplines features, they haven't common data model and related benchmark. This causes lots of geosciences data can't be integrated and shared web sharing oriented. Faced to this problem, the paper analyzed geosciences data multi-dimensions features, including their disciplinary dimension, thematic dimension, spatial scale dimension, temporal dimension, application dimension, etc. Geosciences data concept model with common dimensions was designed based on case study on Data Sharing Network of Earth System Science (DSNESS), which is one of important platforms of SDSP and NSTI. Three common data benchmark of this model are researched based on the geosciences data in DSNESS, i.e., data elements dimension benchmark, spatial dimension and temporal dimension benchmark. Element dimension focuses on the element classification structure and coding system, spatial dimension is contribute to the construction of data spatial reference based on grid, temporal dimension is helpful for the construction of time serials data sets according to the frequency of data updating. The common data benchmark is used in DSNESS and has been proved useful for geosciences data integration and sharing initially.