Geoprocessing Data Types

Thematic Maps

Thematic maps have geographical regions defined by one or more polygons. Examples of such maps are soil usage and the region agricultural capability. These data are obtained by a field survey and they are inserted in the system either by a digitization process or, automatically, from a classified image.

To allow a better representation and analysis of the geographical space, most systems store these map types as arcs (region boundaries), including the nodes (arcs intersection points) to build a topological representation. The topology built has the arc-node-region type: arcs are connected among themselves through the nodes (starting and ending points) and arcs surrounding a certain area define a polygon (region).

A thematic map can also be stored in the raster format. In this case, the area covered by the map is divided into cells with a fixed size. Each cell has a value corresponding to the most frequent theme at that spatial location.

Selecting between the vector or raster representation for a thematic map depends on the map's main purpose. In order to get charts and for operations requiring higher precision, the vector representation should be used. The raster format is the option when the purpose is to perform algebraic map operations. However, for the same precision level, the storage space required by the raster format is considerably higher, this is shown in the figure below.

geo3.gif - 6238 Bytes

The table below shows a comparison among advantages and disadvantages for the raster and vector storage representation for thematic maps. This comparison takes into account several aspects: spatial relationships, analysis, storage. The table highlights the best format in each case.

COMPARISON BETWEEN FORMATS FOR THEMATIC MAPS
Aspect Vector Format Raster Format
Spatial relationships among objectsTopological relationships among available objectsSpatial relationships must be inferred
Database connectionFacilitates the association among attributes and graphical elementsAttributes are associated only to map classes
Analysis, Simulation and ModelingIndirect representation for continuous phenomenon
Map's algebra is limited
This representation is better for phenomena with space continuous variations
Simulation and modeling are easier
Working scalesAdequate for both small and large scalesMore suitable for small scales (1:25.000 and smaller)
AlgorithmsProblems with geometrical errorsProcessing is faster and more efficient.
Storage By coordinates (more efficient) By matrices

Geoprocessing Data Types



Cadastral Maps (object maps)

A cadastral map differs from a thematic map in the sense that in the cadastral map each element is a geographical object, which has attributes and can be associated to several graphical representations. For instance, lots in a city are elements in the geographical space having attributes (owner, location, value, tax, etc) which might have different graphical representation in maps with distinct scales.

The graphical part in cadastral maps is stored using vector coordinates, with the associated topology. It is not usual to represent these data in the raster format.

Geo4.gif - 8776 Bytes

Geoprocessing Data Types

Networks

In geoprocessing the network concept is related to information associated to:

  • Public services, such as water, electrical and telephone network systems;
  • Drainage networks (hydrographic bays);
  • Highways.

In the networks case each geographical object (e.g.:telephone cable, water pipe, electrical transformer) has a precise geographical location and it is always associated to descriptive attributes, presented in the database.

The networks graphical information are stored using vector coordinates, with an arc-node topology: arcs have a flow direction and nodes have attributes (can either be a source or a sink). The network topology is a graph that stores information about resources flowing between geographical locations, as shown in the figure below.

Geo5.gif - 3289 Bytes

As observed by Goodchild (1992b), a network is a 1-D addressing system embedded in the 2-D space. To mention an example, take an electrical network, that has, among others, the components: poles, transformers, sub-stations, transmission lines and switches. The transmission lines will be topologically represented as arcs in an oriented graph, all other information will be concentrated in the nodes. Notice that the algorithms for computing the network properties can, in their majority, be solved just using the network topology and their attributes.

The networks build a chapter aside in the GISs typology, because - the difference for the other data types - are a direct result of human intervention on the environment. Each network application has its own characteristics with a high cultural dependency (e.g. the highway width in the United States is different than the ones in Brazil).

In the networks application case, the connection with a database is fundamental. As the spatial data has relatively simple formats, the majority of the task is related to database consultation, presenting the results in an adequate way. The network area is still a challenge for GIS innovations, mainly:

  • data integration
  • dynamic segmentation
  • visualization language
  • adaptation capability

The data integration is required for applications such as networks, where it is desired to generate a continuous cartographic base from the information spread around several maps. Usually, the networks (electrical, telephone, and water) are interconnected all around the urban mesh. Few systems are capable to store them continuously, originating partitioning that do not reflect the reality and hardening the analysis and simulations execution.

Another required aspect for network applications is the capability to define different logical cuts in a network without having to duplicate (or repeat) the network topological structure. For instance, when a dirt road is partially paved, this information has to be updated without having to retype all the road coordinates locations. This capability, usually denoted as dynamic segmentation, allows to separate the different information levels related to the same network.

As the network element characteristics are stored as attributes in databases, it is required to have some ways to visualize these information. For this, the GISs have to have a presentation language that allows to control the symbolism associated to the network components, which changes with the plotting scale used.

The minimum package available in commercial systems typically consists in the optimal and critical path computation. This basic package is not enough for most applications because each user has completely distinct requirements. In the telephone system case an interesting question might be: "what are all telephone numbers served by this terminal box?" while in a water network system, the question might be: "if a certain chlorine percentage is added to the neighborhood water reservoir, what would be the house's final chlorine concentration?".

In this way, a network modeling system is useful for a client after properly adapted for his requirements, which can take years. This adds a basic characteristic for this application: the systems have to be versatile and flexible.

Geoprocessing Data Types

Images

Images can be obtained from satellites, aerial photography, or satellite scanners, the images represent indirect capture ways of spatial information. Each image element (named pixel), stored as matrices, has a proportional value to the soil reflectance in the considered area.

imagem_tm_manaus.gif - 35467 Bytes


By the image acquisition process nature, the geographical objects are contained in the image and to individualize them, it is required to use photo-interpretation techniques and automatic classification.

Important satellite image characteristics are: the number of bands in the imaged electromagnetic spectrum (spectral resolution), the instantly observed earth surface area by each sensor (spatial resolution), and the interval between two consecutive satellite passages over the same point (temporal resolution). The table presented below shows the general characteristics for the main satellites (and respective sensors) available in Brazil.

CHARACTERISTICS OF SATELLITES AVAILABLE IN BRAZIL
Satellite (family)InstrumentBands NumberSpatial ResolutionTemporal Resolution
LANDSATMSS 4 80 m18 days
TM 7(1) 30 m18 days
SPOTXS 320 m27 days(2)
PAN 1 10 m27 days
TIROS/NOAAAVHRR 5 1100 m 6 hours
METEOSATMSS 4 8000 m30 minutes
ERS(3) SAR band-C 1 25 m25 days

 

Notes:
1. The TM sensor has a thermal infrared band with a 120 m resolution.
2. The SPOT satellite has a moving mirror with lateral view, allowing a reduced spacing for two consecutive shots of the same Earth point.
3. The ERS-1 satellite is the first one to have a radar sensor. However, it is an experimental satellite with limited coverage.

Geoprocessing Data Types

Digital Terrain Models

The Digital Terrain Model term (or DTM) is used to denote a measurement that continuously changes in space. Usually it is associated to altimetry, also it can be used to model geological units, such as mineral or soil/subsoil properties rates (such as aeromagnetic properties).

Among several digital terrain model applications, the most commons are (Burrough, 1986):

  1. Altimetric data for topographical map generation;
  2. Earthwork analysis for highways or railroads projects;
  3. Declivity map and exposition computations for geomorphologic and erodibility analyses and surveys;
  4. Geophysics and geochemistry variable analysis;
  5. Tridimensional presentation (usually combined with other variables).
The acquisition process for a spatial variation value usually produces a set of point samples. From these samples, it is possible to build two representation types:
a)Regular grids: the element matrix has a fixed spacing, where the value associated to the grid point is estimated. The regular grids are obtained by samples interpolation or, alternatively, generated by restorers with digital output.
b)Triangular grids: the grid is formed by a connection among samples, using the Delauney triangulation (with some constraints). The triangular grid is a vector topological structure of the arc-node type, forming a set of irregular carvings in space.

Geo6.gif - 5361 Bytes

The interpolation procedures for regular grid generations from samples changes with the measurement. In the altimetric case it is common to use the weighted functions by the inverse of the squared distance. For geophysical variables, bidimensional filtering procedures or geostatistics (Kriging) are used.

The triangular grids are usually better for terrain variation representation, because it captures the relief complexity without requiring a large amount of redundant data. The regular grids has large redundancy in uniform terrains and they are hard to adapt to complex relief in the same map, because the sampling spacing is fixed.

For the geophysics variables and for 3D visualization operations, the regular grids are preferable, mainly because it is computationally easier to handle it.

The table below summarizes the main advantages and disadvantages of regular and triangular grids.

COMPARISON BETWEEN REGULAR AND TRIANGULAR GRIDS TO REPRESENT DIGITAL TERRAIN MODELS
Triangular gridRegular grid
Advantages 1. Better representation for complex reliefs.
2. Incorporates constraints such as ridge lines
1. Facilitates handling and conversions.
2. Adequate for geophysical and 3D visualization
Problems1. Complex to handle.
2. Inadequate for 3D visualization
1. Complex relief representation.
2. Declivity computation

The digital terrain models can also be converted into thematic maps or images. In both cases, the numerical value is quantified, either for a small number of values (in case of thematic maps) or for variations associated to images (discrete values).

Geoprocessing Data Types