Embedded GIS for census mapping

A. Sai Venkata Lakshmi , P. S. Bhavani Kumar, J. Sai Baba
ADRIN Secunderabad - 500009,India
Tel:91-040-7754926, Fax: 91-040-7754978
asvl@yahoo.com

Abstract
This paper discusses the development of a utility for visualizing spatial data such as census using embedded GIS technology. We present an efficient, cost-effective method that utilizes the best features of databases and GIS packages: efficient data storage and retrieval as well as viewing the data spatially so the underlying pattern can be observed. The solution is to store the non-spatial data in a database and relate it to a spatial layer which is stored in a GIS supported format like shapefiles The user first defines a relation between a field in the database and a field in the spatial layer. The user can perform a number of operations like zooming, panning and searching for features in the spatial layer. One can dynamically query the database and view the corresponding spatial components. The data may reside anywhere on a network and may be stored in any database that is ODBC compliant. The spatial data can be visualized in a number of ways that enhance the user's understanding and interpretation of the data, like comparing multiple attributes of a feature by depicting the attributes as elements of pie charts or bar charts. In addition, one can compare one feature to another by the relative size of each feature's chart. Features can also be classified based on attributes or symbolized with unique attribute values. Polygons can be filled with a dot density proportional to attribute values. This is an ideal solution for users who have database applications and want to add a spatial component with a few GIS operations.

Introduction
The proliferation of GIS and desktop mapping packages in the last decade has brought GIS technology to the desktops of citizens, planners, decision makers and administrators. When we look at major forces of the 21st century like population growth, economic development and consumption of natural resources, we see that all of them have a spatial component. Many corporates store large volumes of data in an RDBMS for quick retrieval and easy maintenance. Data produced by Census is a primary source of information needed for effective development, planning and monitoring of population, socio-economic and environmental trends. Though stored in databases, this data has a spatial component inherently associated with it. The census database contains a lot of attribute information which can be linked to spatial units by spatial referencing. Relating the spatial component along with the non-spatial attributes of the existing corporate data enhances user's understanding and gives new insights into the patterns and relationships in the data that would otherwise not be found.

Embedded GIS
As the software industry is becoming component-based, the GIS industry is providing components that can be embedded in other applications. The evolving standards for software components such as ActiveX, COM and Javabeans, as well as industry specific Open GIS Consortium standards are making it possible to embed GIS capabilities in mainstream IT applications. ESRI's Mapobjects, Mapinfo's MapX and InterGraph's GeoMedia objects are examples of products that provide GIS functionality in individual, accessible software components. GIS software is easy to use, less expensive and seamlessly integrated with standard word processing, spreadsheet and database applications.

Using Census Mapper to visualize Census data
The objective is to enable viewing of existing census data stored in an RDBMS which has an inherent spatial component. We present an efficient, cost-effective method that utilizes the best features of databases and GIS packages: efficient data storage and retrieval as well as viewing the data spatially so the underlying pattern can be observed. The solution is to store the non-spatial data in a database and relate it to the corresponding spatial layer which is stored in a GIS supported format like shapefile.

The user first defines a relation between a field in a table in the database and a field in the corresponding spatial layer. Any number of relates can be made. For example, if there is a table that holds state-level data which has a filed 'STATENAME", then a relate is made on this field to the field 'Name' in the spatial layer 'states'. The table containing the district-level data can correspondingly be related to the spatial layer holding districts. A similar relate can be made at the city-level.


Diagram showing program flow

If the data is not enumerated or a many-to-one relationship exists between the database and spatial layer , a new table can be created by appropriately summarizing the data and this table can be related to this spatial layer. Otherwise code can be included to convert the many-to-one relationship to a one-to-one relationship appropriately. The database is queried and the selected records are passed to the application, which selects the corresponding spatial objects. This enables dynamic mapping, i.e. what you see on the map changes based on other related factors.

Features
Various features like panning, zooming, searching for features satisfying a criteria and proximity analysis are available. This provides a framework for viewing, analyzing and supporting decision making on different scales. The information is first viewed at the state level, further zooming displays the district-level data. Zooming onto a district then gives town level data, wherever available. Thus one can view the variation between states, the variation between different districts within a state along with the variation among towns in a district. This is illustrated in Figures 1 to 3, where six fields Total Male Population, Total Female Population, Total Male Literates, Total Female Literates, Total Male Workers and Total Female Workers are represented as bar charts. Fig 1 shows data at state level, Fig2 at district level and Fig.3 at town level. One can also flash features that are selected based on some criteria.

Visualizing enumerated data aids planners and decision makers. Symbols positioned to represent sample locations can be visually configured in order to display recorded attributes to enable geographic patterns to be interpreted at different scales, and the detail for local variations to be detected and data values perceived. The spatial data can be visualized in a number of ways that enhance the user's understanding and interpretation of the data, some of which are mentioned below. One can choose which fields to use for the charts. One can compare multiple attributes of a feature by depicting the attributes as elements of pie charts or bar charts. One can also scale the bar and pie charts based on a field like population. In addition, one can compare one feature to another by the relative size of each feature's chart. Some illustrations of bar charts are shown in figures 1 to 3.To simplify visualization, attributes or density values can be classified into a few categories that span the data range. Each category can be assigned a shade in a graded sequence of a user-defined color ramp. This is illustrated in Fig 4, where the states of India are coloured based on total population. Features can also be symbolized with unique attribute values. Areal units can be filled with a dot density proportional to attribute values.