22 Geo-relational and object oriented data structure
Dr Dinesh Kumar
1. Learning Objectives: To understand the concept of Geo-relational and object oriented data structure in
GIS.
2. Introduction:
Spatial data is any data pertaining to space. Thus, for storing, analysing and retrieving spatial data, it needs to be imported to GIS platform. The data models may be vector (represented by points, lines or polygons) or raster (represented by pixels). Such representation of spatial data in either vector or raster format in GIS platform may have its own advantages and disadvantages, but are essential for carrying out analysis of spatial data. The spatial and non-spatial information are stored within GIS platform in a structured or organized way. Geometrical shapes along with its attribute should be in synchronized way in order to represent the spatial scenario. Storing spatial geometry, spatial relationships among the spatial feature and their attributes makes it challenging task. Since, spatial and non-spatial data acts as base for all the analysis, queries and processing, data management in a structured or organized way becomes crucial for developers. There are a number of such database management system which provides efficient, quick, and conditional and space saving capability within the GIS platform. The analytical and modeling capabilities of GIS make it ideally suited for solving complex management problems and helping in better decision making.
Data model should meet the following criteria
2.1. Generality:
Data model should have feature to support a variety of data bases and suitable for range of application such as thematic mapping, land inventory and topographic mapping.
2.2. Simplicity:
Simplicity decides the efficiency and reliability of geographic data and algorithms. It should be as simple to meet the expectations.
2.3. Efficiency:
The data model should implement geo-processing tool efficiently without conversion of data into particular format.
2.4. Adaptability:
The data model should meet the requirements of simple user to system programmer.
2.5. Freedom from Restrictions:
The model should be free from most of the limitation in terms of size or content. Handling of small to large dataset should be well.
Over the period of time a number data model were developed. Among them geo-relational and object oriented data structure is discussed here after in detail.
3.1. Geo-relational Data Structure
Geo-relational data structure stores vector data for a geographic coverage which includes it’s spatial and attribute component. Spatial data describes the location of spatial features, whereas attribute data describes the characteristic of spatial features. Geo-relational data structure organizes the data into split system which means it stores
- Spatial data (Geo) into graphic files
- And Attribute data into relational data base
As we know that the vector data uses geometric objects of point, line and polygon to represent spatial feature. Spatial representation of spatial feature is complex as their (point, line and polygon) relations on earth are not simple. Therefore, topology acts as set of rules to define the interrelationships among spatial feature. Coverage does also posses the same feature to define rules to define the interrelationships among spatial feature. There are packages embedded within GIS platform to define spatial set of rules among the geometrical objects.
Now it becomes clear that geo-relational data structure requires the following in order maintain data integrity and consistency.
- Spatial data (Geo) into graphic files
- And Attribute data into relational data base
- Coverage or topological set of rules
Graphic file, relational database and coverage or topological files, assigns common ID (Identifier) to particular feature class and it’s associated topological and attributes files. This identifier acts as identity for that particular feature and its association with other spatial feature. Using the common ID data can be synchronized and processes like query, analysis and display in unison can be performed.
Most of the GIS platforms are equipped with geo-relational data structure.
Fig-1: Geo-relational data Structure
3.2. Object Oriented Data Structure:
The Object oriented data structure is designed using relational DBMS with object oriented extensions. This data structure has robust management capabilities of RDBMS along with access of object oriented system. Complex column structures can be defined using object oriented and RDBMS. Object oriented data model stores spatial and attribute data together rather than in a split system.
This model also allows spatial feature (object) to be associated with properties and methods where proper is an attribute or characteristic of an object and Method –is a specific action that can be performed on an object.
Object Oriented Data Model stores data at two separate layers, namely geographic and geometric object layer. Generally object data model at higher level is known as geographic object data model. The geographic objects, virtual objects and a set of semantic spatial functions, retrieval functions, set functions, and aggregate functions are termed as geographic object data. It also has a set of functions for retrieval, manipulation, and computation of geometric objects. Lower level data model is known as geometric object data model which consists of primitive, geometric and virtual objects. Geometric objects are actual spatial representations of the geographic objects such as polygons, nodes and edges. Spatial computation creates geometric objects. Object Oriented Functional Query Language (OFQL) interface for the database access has been embedded within the system for the more complex calculations. The model provides abstraction and functional independence between the geographic and geometric object data model. Each model has its own set of object classes, objects and functions. The parent and child relationship within the model is represented by PARTOF relationship and IS-A abstraction. The model takes care of the PART-OF relationship and IS-A relationship by introducing various functions for resolving the containment topological relationship.
Fig-2: Object-Oriented Data Model
Object based data model is built on four basic concepts of abstractions namely
- Classification
- Generalization
- Association
- Aggregation
3.2.1. Classification:
It is the process to map objects (instances) onto a common class. Object is defined as a single occurrence (instantiation) of data about individuality and observable behavior.
The object oriented approach, every object is considered as an instance of class. Therefore classification is also termed as instance_of relationship. And type defines the behavior of its instances by operators which manipulates the objects.
Fig-3: Graphical representation of classes. Source: Object oriented modeling of GIS, Max J. Egenhofer and Andrew U Frank. 1992
3.2.2. Generalization:
This combines many classes of objects with common operators into general super class. Super class defines the grouping and its relation with subclass. Subclass and super class is abstraction of same class. Further super class may have multiple subclasses and at multiple levels, subclass may act as super class for some classes..
3.2.3. Association:
It defines the relationships among independent objects which is met by higher level of set objects. Set defines the association and members (associated objects). Therefore this abstraction process is termed as member-of relation.
3.2.4. Aggregation:
This explains about composed objects which mean that a group of object may form a semantically higher level object and is termed as aggregate or composite objects. Aggregate operation is not compatible with part operation. The relation defined by aggregation is termed as part-of relation. Aggregation applied to components generates aggregate type data structure.
Fig-5: A city model as aggregate. Source: Object oriented modeling of GIS, Max J. Egenhofer and Andrew U Frank. 1992
Inheritance and propagation are very crucial while modeling the complex object behavior. Object oriented data modeling can solve the purpose of suitable spatial information system. This modeling approach can be applied at all the stage of large complex software system. This approach particularly abstract mechanism is necessary to model the complex situation such geometric objects.
4. Conclusion
The spatial and non-spatial information are stored within GIS platform in a structured or organized way. Geometrical shapes along with its attribute should be in synchronized way in order to represent the spatial scenario. Since, spatial and non-spatial data acts as base for all the analysis, queries and processing, data management in a structured or organized way becomes crucial for developers. There are a number of such database management system which provides efficient, quick, and conditional and space saving capability within the GIS platform. The Geo-relational data structure store the data in a split system while object oriented data model store the data in a single system. Object-oriented data got upper hand in synchronizing the spatial and non-spatial data along with its spatial relationships.
Suggested Readings
- http://www.geoinformatie.nl/courses/grs10306/Wenting/WYSIWYG/WYSIWYG_in_context. pdf
- http://shodhganga.inflibnet.ac.in/bitstream/10603/110137/11/10_chapter%2002.pdf
- http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.473.2688&rep=rep1&type=pdf The McGraw Hill Companies, Inc.
- Object oriented modeling of GIS, Max J. Egenhofer and Andrew U Frank. 1992