The polymorphic, diversified, multi-sources and multi-use scientific data is quite different from the other data, there are many new challenges in building Scientific Data Management and Analysis System (SDMAS), especial one is that the multi-use data requires alternate classifications and mappings between them. Based on the oceanographic data, we present the solutions of multi-classifications problems. We first introduced the data and classificatory criterions of this case, then adopt the classification ontology to represent multi-classifications, use the metadata, query builder and data view mechanisms to support the dynamic cross-classification oriented query. We also prove the purposed solutions by applications and experiments. Theories and practices prove that the proposed solutions are contributed to build multi-classification based SDMAS.