Abstract: Through a dialectic examination of basic properties of space and time,such as independence and relevance,absoluteness and relativeness,dynamics and statics,the geographical spatio-temporal data model is generally regarded as a profile of the temporal geographical information system.In this way,the various spatio-temporal data models are conceptually generalized.Also,novel improvements of nested cube,hierarchical composition,high-order difference,and variable-resolution snapshot on flat-structuralized Gail Langran data models is proposed.Furthermore,a multi-scale spatio-temporal data model is algebraically described.Initially,the formal generalization of Gail Langran's data models and their variants is completed.Undoubtedly,this work will greatly speed up the development of scalable and self-adaptive temporal geographical information systems.