A time series data management method for geographic information systems

摘要

The need to manage time series data is increasing in geographic information systems (GIS). GIS records the changes in the real world (for example, houses built, roads constructed), and occasionally spatial calculation on an old time is requested. Other than the time series data, spatial relation is absolutely necessary in GIS. Spatiotemporal data management is also requested. Most daily work using GIS is based on the present data. In order to keep the access speed up the present data must be paid the most attention, even in spatiotemporal GIS. The access speed for the present data should not be lowered by adding the capability to handle temporal data. In addition, past data should be able to be restored. This paper proposes a new spatiotemporal data management method based on a hierarchical spatial data structure (GBD-tree) and the geographic differential script file (GDSF) method. The GDSF is a method used to record changes in maps in the form of a graphical operation (e.g.insertion or deletion of an entity) with a time print. More precisely, the GDSF records the change from the present to the past. When old data becornes necessary, the data is restored by applying the GDSF to the current map data while the time print is newer than the specified time. Since the proposed method is designed so that the restoring operation is executed in a buffer during a spatial retrieval, this operation has no effect on the present data. This data structure does not require extra time to retrieve the present data. On the other hand, the time consumption to restore the old time data is proportional to the length of time from the specified time to the present. This paper describes the data structure and the retrieval algorithms.