Data Capture

Data input to a geographical information system can be best broken into three categories: entering the spatial data, entering non-spatial data, and linking the two together.  Entering the spatial data can be done numerous ways.  Spatial data can be acquired from existing data in digital or paper form, or it can be collected from scratch.

Finding already mapped data in a paper format for an area can be accomplished in several ways.  Paper map collections can usually be found within large libraries or universities.  Libraries often times will also contain books with maps for international and domestic data.  Another good resource for geographic data is local, state, or national government.  Many countries have a wide range of data available at their country mapping agencies.  If the data is to be more localized to a specific area, the local governments such as planning departments should contain the information.  In addition, there are many commercial mapping companies that will sell data world wide for certain countries.  The Internet is a good resource to search for data either from a vendor or a site offering free data. (Clarke, 2001)

There are two methods of getting paper maps into the computer: digitizing and scanning.  Geocoding is the term used for the conversion of analog spatial information into digital form.  Digitizing on a tablet captures map data by tracing lines by hand, using a cursor and an electronically sensitive tablet, resulting in a string of points with (x,y) values.  Scanning involves placing a map on a glass plate while a light beam passes over it, measuring the reflected light intensity.  The result is a grid of pixels.  Image size and resolution are important to scanning.  Small features on the map can drop out if the pixels are too big. (Clarke, 2001)

Finding data via the Internet can be done by performing a basic search.  There are several sources for downloadable data such as:

• The Geography Network
• Data Depot
• Spatial Information Clearinghouse

Finally, if the data available does not meet the needs of the user, it can created by use of GPS, Remote Sensing, Aerial Photography, and field collection techniques.

Projection and Rectification

In order for the spatial data of a 3-dimensional earth to be represented in a 2-dimensional GIS, the data must make use of one of the various projection methods (See Remote Sensing Section for further detail on projections).  Because different projections place the same special entities on different coordinates on the flat surface, it is vital that a projection be set for the specific data set being used.  One of the main features of a GIS is the ability to overlap different data layers for better analysis.  These different layers must have the same projection, datum, and reference ellipsoid so that all coordinates are lined up correctly.

 
Figure 1, Reference Ellipsoid and Geoid. (SIC, 2002)

Data Modeling

Spatial modeling represents the structure and distribution of features in geographical space.  In order to model spatial processes, the interaction between these features must be considered.  There are several types of spatial data models including: vector, raster, surface, and network (Burrough, 1998).

Figure 2, Integrated Layers of GIS Model (SIC, 2002)  

Vector Data Model

The vector data model is a method of storing and representing data on an X,Y Cartesian plane.  A coordinate and an equation defining the curvature of each feature is stored for both the beginning and the end point of each feature.  The building block of the vector structure is the point; lines and areas are composed of a series of points in a specific order that gives the object direction (Clarke, 2001)  The attribute data in the vector model is stored in a separate table that can be linked to the map.  Because every item on the map has its own separate attribute data, analysis can be very easy.  For example, if a vector road network is being used to analyze the amount of carbon monoxide produced by cars per year in both rural and urban communities, each road would be capable of having separate attributes, thus allowing the GIS user to view or select each road and access information associated with just that road.

Vector data entities in a GIS hold individual values, for example, if two lines overlap, unique values are recorded for each line in the database (spaghetti model).  Selecting an appropriate number of points is another consideration to be made with vector data; if too few points are chosen, the shape and properties of the entity will be compromised and if too many points are used, duplicated information can be stored resulting in data overload (Burrough, 1998)

Figure 3, Vector Spaghetti Model (SIC, 2002)

Raster Data Model

The raster data model uses a grid composed of rows and columns to display map entities.  Each cell in the grid is equivalent to one map unit or one pixel.  Spatial resolution determines the precision of spatial representation by raster data. The smaller the size of the pixel, the higher the resolution and the better the precision of spatial representation (Lo, 2002).  An entity code is assigned to each cell that is connected to a separate attribute table, which provides information to the user as to what entity is present in what cell.

Figure 4, Raster Representation(SIC, 2002)

 
Figure 5, Raster Attribute Table(SIC, 2002)

The term raster data when applied to GIS and mapping includes scanned monochrome and color printing separates, scanned black and white and color aerial photographs, remote sensing images, digital elevation models, as well as thematic spatial data created by manual and computer-based methods (Lo, 2002).  These methods of storing one or more values for each grid location in the data drastically increase the file size.  Several methods have been developed to compact the size of raster files.  The first is run length encoding which reduces data on a row-by-row basis.  If an entity occupies a large number of cellsin a row, a single value is stored representing the object followed by the number of cells in that row, rather than recording each individual value.  Another compaction technique is called the quadtree data model. In this model, instead of dividing the entire area into cells of equal size, only areas with specific details are broken down into smaller cells.  For example, if a land-use map had only one land use type, one cell would represent the entire area.  If there were 4 classes, 4 cells would be used, and quadrant that had more than one land use type would be broken down until it only contained one type (Lo, 2002).

Figure 6, Quadtree Compaction (SIC, 2002) 

Figure 7, Run Length Encoding(SIC, 2002)  

The raster data model represents spatial phenomenon such as topography, land use cover, and air quality as categorical or continuous surfaces.  This makes raster-based methods particularly suitable for spatial modeling that involves multiple surface data sets.  However, this method is not suitable for applications that rely on individual spatial features represented by points, lines, and polygons (Lo, 2002).

Tabular Data

Tabular data, also called attribute or descriptive data, is one of the most important elements in a GIS.  It is statistical, numerical, or characteristic information that can be attributed to spatial features.  Similar to spatial data the tabular data is stored by the GIS software in a method that allows it to be accessed and viewed, usually in a relational database format.  Depending on the application, attributes that may be useful to assign to a feature would be population of an area, traffic measurement of a road, or types of landmines in a particular area.  The GIS software allows the attribute data to be linked to the spatial data in such a way that it gives the attributes a location.  A GIS package knows a specific location geographically from the storage of spatial data.  By linking attribute data to the spatial data, the GIS package knows some of the characteristics of a feature in the spatial data set.

Two or more tabular databases can be linked when there is a common data filed.  This allows the GIS to become a powerful spatial analysis tool.  A GIS user, after integrating both spatial and attribute data, has the capability to learn a great deal about the defined study area.