Developments of global climate observation system (GCOS) greatly benefit atmosphere composition research. The observation data from GCOS makes it possible to generate essential climate variables (ECVs) of different categories. Data with various spatial and temporal resolutions paves way for long-term atmospheric observations under different spatial scales. However, difficulties in data discovery and sharing served as obstacles for leveraging GCOS data in atmosphere composition researches. Specifically, lacking standard keywords describing atmosphere composition impedes the data discovery in broader categories. Lacking geospatial catalog capable of serving data from different agencies restricts data discoveries in broader scopes. To tackle aforementioned challenges, a geospatial data discovery framework based on GCMD science keyword system and federated geospatial catalog is proposed. First of all, dataset is discovered with atmospheric composition keyword through GCMD. Secondly, granules under discovered datasets are searched through federated geospatial catalog. The federated geospatial catalog is responsible to interact with heterogeneous data providers. Summarily, on one hand, proposed framework facilitates discovering atmosphere data with different categories. On the other hand, the federated geospatial catalog broadens the scope of atmospheric data discovery.