Temporal data management is an extended field of data management, and the research on this field has theoretical significance and practical application value. The index of temporal data is crucial to temporal data management and thus is one of the hot research topics. First, based on the partial-order mathematical relationship, this paper proposes a data structure which can convert a temporal query processing of two-dimension into one-dimension. This structure can process all the temporal relationship operations efficiently. Secondly, this paper presents an indexing scheme called TempPartialIndex which combines temporal partial-order data structure into non-temporal XML index. In TempPartialIndex schema, the processing of query begins with mapping and filtering with the semantic and temporal constraints so that a large number of nodes will be filtered out and only a few nodes will be left. Then a structural join algorithm is executed on the left nodes. Furthermore, the paper discusses the query based on set-at-a-time, the temporal variable query and modification. Simulations show that TempPartialIndex can process the temporal XML query and update efficiently.