This paper analyzed the features and limitations of several typical spatio-temporal data models. "spatio-temporal cube":
the main disadvantage is that the target change will produce great data redundancy when under non-consecutive
circumstances. "Snapshots": it repeatedly saves graphics and attribute of no changes which resulted in waste of storage
spaces, and it is impossible to reflect space objects under the same domain and the relationship between the attributes.
"Base State with Amendments": merely modify changing object, but it's not suitable for continuous variation space
object. "space-frame composite": currently, the model is lacking of sound framework structure and application model.
"Object-oriented spatio-temporal model": The modeling concept, theoretical foundation and technical realization has not
yet reached a consensus, it's not mature enough.
In allusion to the features of the spatial database of water and soil loss, this essay expounded the characteristics of spatiotemporal
databases. Spatial features in many practical circumstances ( such as thematic maps in soil and water
conservation projects and space elements of soil erosion distribution map) have spatial data features, and also change
with time, consequently, required us to establish spatio-temporal database, STDB, which can capture time data and space
data at the same time. This analysis based on "ArcSDE versioning mechanisms" temporal and spatial database
implement technologies, discussed the construction methods, process and data features of the database, and introduced
the implementation of historical data rebuilding and version merging.