WORKING WITH SPATIO-TEMPORAL DATA TYPE

Several aspects of spatiotemporal databases have been explored in past decades, ranging from basic data structure to query processing and indexing. But today, operational temporal GIS does not exist. The key impediments have been the complexity of integrating space and time and the lack of standards. OpenGIS standards for simple feature access (spatial type) do exist, but unlike the spatial type, standards for spatiotemporal data type do not exist. This paper explores a new approach to modeling space and time to provide the basis for implementing a temporal GIS. This approach is based on the concept of data types in databases. A data type provides constructors, accessors, and operators. Most commercial and open source databases provide data types to deal with the spatial component of a GIS, called spatial type. Oracle Spatial, DB2 Spatial Extender and Informix Spatial DataBlade, ST_Geometry for PostgreSQL and Oracle from Esri, PostGIS for PostgreSQL, etc., are some examples. This new spatiotemporal data type is called spatiotemporal type (STT). This STT is implemented in PostgreSQL, an open source relational database management system. The STT is an extension of Esri's ST_Geometry type for PostgreSQL, in which each spatial object has a life span. Constructors, accessors, and relational functions are provided to create STT and support spatial, spatiotemporal, and temporal queries. Some functions are extended based on OpenGIS standards for the spatial type. Examples are provided to demonstrate the application of these functions. The paper concludes with limitations and challenges of implementing STT.


INTRODUCTION
Discussion on spatiotemporal databases is not new.Several aspects of spatiotemporal databases have been explored in the last two decades.There is a bewildering array of research ranging from simple/complex data structure to query language and spatiotemporal indexes.The space-time cube by Langran (1992), object-oriented approach to handle spatiotemporal objects by Worboys (1992), and cell tuple-based spatiotemporal data model by Raza and Kainz (1999) are major examples of data models or structures.Spatiotemporal query language by Snodgrass (1993) and spatiotemporal indexing techniques by Jensen et al. (2006) and Hadjieleftheriou et al., (2006) are other research examples.Detailed discussion on these researches is beyond the scope of this paper.
In spite of these developments, we do not see any operational temporal GIS.The key impediments have been attributed to the complexity of integrating space and time and the lack of standards for spatiotemporal data.Software standards ensure desirable characteristics of software.It ensures that it is implemented uniformly across the platforms and software inputs/outputs are known.OpenGIS standards for simple feature access (spatial type) do exist (OpenGIS, 2010) but unlike spatial type, standards for spatiotemporal data type do not exist.Many commercial and open source software conform to these standards.This paper presents a new approach to modeling space and time to provide the basis for implementing a temporal GIS.This new approach is presented in section 2. Section 3 discusses various constructors, accessors, and operators for this new data type.Examples are provided in section 4. Limitations and further work are discussed in Section 5.

SPATIOTEMPORAL DATA TYPE
GIS is becoming an integral part of any enterprise database and no longer considered a separate branch from mainstream information technology (IT).As part of a standard database management system, a GIS is supposed to have a data type like any other data type in any database management system.This paper introduces a new data type for space-time modeling.
A spatiotemporal data type is a data type where space and time is integrated and time is not considered an attribute of space.A similar approach was presented by Erwig et al. (1999) for moving objects.Basic objects in that approach were evolving points and regions.Line objects are considered as projections of movement.This approach is good for moving objects such as cars or airplanes but may not be best for urban applications such as land-use change or tracking parcel history where the object does not move rapidly.Moreover, it is not clear how to model line features such as roads in that approach.
The data model for STT is defined by an object model.In an object model, everything is an object.A class is a set of objects with common properties.A class consists of data members and member functions (operations).The structure of the data model is defined by data members (data), while operators and consistency rules are defined through operations.
Data structure and consistency rules for the model are discussed in this paper.The basic element of this model is a spatiotemporal object.This spatiotemporal class is the aggregation of spatial and temporal classes.The spatial and temporal object and associated operations are well understood in the respective domains and are not discussed here.This paper focuses on integration of spatial and temporal class.The attributes and operations of a spatiotemporal class (STobject) are described in figure 2. The operations can further be classified into three main categories: constructors, accessors, and operators.These are discussed in the next section.
In figure 2, srid is spatial reference system ID; st_entity_type is an entity type such as STPoint or STPolygon; minx, miny, maxx, and maxy are minimum and maximum coordinates of the spatiotemporal object; ts and te are start and end times, respectively; and tshape is the binary internal representation of the spatial part of a spatiotemporal object.
The schema for ST-type is presented in figure 3. The Feature schema has feature column type_column and other properties.
The ST_Type schema contains information about the Feature schema and srid.The Spatial_References and Coordinate_Systems schemata store the information about srid.

CONSTRUCTORS, ACCESSORS, AND OPERATORS
As mentioned earlier, a type consists of constructor, accessor, and operator functions.A constructor is used to create a type.Two types of constructors are provided: - The first two formats (#1 and #2) represent time instance (T0), while the rest of the format shows the time interval (T1 Then, the operation on st_type can be defined as Where pure spatial relation as The spatiotemporal STT_time_overlaps can be defined as Describing all functions is beyond the scope of this paper.These above functions are shown to demonstrate the difference between spatial and spatiotemporal operators. The spatiotemporal data type is implemented in PostgreSQL 9. 1.2 (PostgreSQL, 2011).The STT is an extension of Esri's ST_Geometry type for PostgreSQL.Section 4 describes some implementation examples.

CONCLUSION
The paper presents some initial results of ongoing research in spatiotemporal data type.This research is still in its early stage.
A new data type is presented in this paper to deal with complex issues of space-time modeling.The list of operator, constructor, and accessor functions is not complete and more work is needed to document remaining functions.Currently, binary data is returned in WKB or Esri shape format (data without time).
There is a need to return binary data in a format where time is also stored.Future work involves indexing this data type.Without indexing, performance cannot be achieved.Full implementation of this data type would help us implement temporal GIS.

Figure 1
Figure 1 depicts two classes: spatial and temporal.The object of a spatial class is a spatial object in Euclidean space while a temporal class represents the linear time.

Figure 1 .
Figure 1.Spatialtemporal class Only valid time called WorldTime (WT) is considered in this implementation.Figure 1 is the simplified ST data model.Each spatiotemporal object, STPolygon, STMultiPolygon, STLines, STMultiLines, STPoint, and STMultiPoint, has start (T Start ) and end (T End ) times.

Table 2
Table 1 illustrates temporal relations.STTF_time_overlaps is a special case where oredering is ignored.Pure spatial and temporal relations start with the prefix STTF_, while spatiotemporal relations start with the prefix STT_.Table 2 lists some of these operators.
). 'NOW' represents the open time, which means the object is still alive.At this time, only the date format is supported, time is not included.The binary format can take WKB (Well Spatial and temporal are regular operators employed in spatial and temporal databases.Temporal relations are analogs to spatial topological relations but a bit different than spatial topological relations.The former has the sense of direction, or in other words, these relations are ordered topological relations. .Spatiotemporal operators Operators in table 2 can formally be defined as follows: let A and B are two st_type objects, g is geometry, t is time, α is boundaries, β is interior, and γ is exterior of st_type object.

Table 3 .
Number of parcels by dateTo demonstrate the concept, let's create a simple parcel table to insert and query data.
-Create table parcel (objectid serial, shape st_type); Constructors can be used to insert st_type into the database.For example, the ST_Type constructors can be used to insert parcel data with time into the parcel table.