![]() Modeling ProcessesThe Real World UniverseThe SPRING model is not limited to a particular Geoprocessing application area but it is capable to incorporate several different applications such as Environmental studies, Agricultural, Geology and Networks. The following table presents a mapping as a function of the data types of all Geoprocessing application areas.
How to create a Project ![]() The Conceptual UniverseIn Geoprocessing, the geographical space is modeled following two complementary views: the field and object models (Worboys, 1995). The field model sees the geographical space as a continuous surface, over which varies the observed phenomena, following different distributions. For example, a vegetation map describes a distribution associating each point in the map to an specific type of vegetation cover, meanwhile a geochemical map associates to each point the contents of a mineral. The object model represents a geographical space as a distinct and identifiable entity collection. For example, the spatial cadastral of land in a town identify each part of land as an individual data with its own attributes, distinguishing it from others. In the same way it is possible to think as geo-objects: the rivers in a hydrographical region or the airports in a state. Geoprocessing applications deal with these two data types:
A fundamental aspect for the distinction between fields and objects is the question related to their identity. There are thousands of areas in Brazil that can be classified as “Dusky-Red Latosol”, but there is only one “Rio de Janeiro Botanical Garden”. A field existence is only materialized when its geographical spatial representation is defined. The geographical region defining a “Dusky-Red Latosol” area in São José dos Campos is not an individual entity. What is identifiable, in this case, is the “Paraíba Valley soil map”. The objects has an independent existence from its map representation; in a Geoprocessing system, the objects are usually created from their own attributes and after it they can be located in the space. For example, it is possible to talk about objects such as “São José dos Campos schools”, and, specifically the “Mater Dei School”. Conceptual Universe ClassesInitially it is important to establish the geometrical base where the model classes are defined. Starting from a continuous region on the earth surface it is possible to define the concept called geographical region (or geographical grid). “We define a geographical region R as a general surface belonging to a geographical space, which can be represented as a plane or as a grid, depending on the appropriate cartographical projection.” The geographical region R is used for geometrical support for geographical locations, because all geographical locations will be represented by a unique point from R. The usage of a discrete set of points facilitates a formal definition of the geographical data class and its associated operations. The geographical region definition proposed does not constrain the geometrical representation choices (vector or raster) associated to the geographical objects. The model basic classes, defined bellow, are: ![]() Representation UniverseWhile discussing the representation universe, it will be pointed out what are the used structures to build up a Geoprocessing system. The basic concept is GEOMETRICAL REPRESENTATION. A representation defines a geometrical description in an information layer, which can be specialized in one of the following:
![]() Representation Universe
Raster Representation HierarchyTHE RASTER GEOMETRICAL REPRESENTATION can be specialized in classes as presented in the figure bellow.
![]() RASTER REPRESENTATION class hierarchy.
Vector Representations HierarchyIn order to define this hierarchy we need first to define better what is understood by geometrical primitives: 2D coordinates, 3D coordinates, 2D node, network node, arcs, oriented arcs, contour lines and polygons. Given a geographical region R, it is possible to define:
Once the vector geometrical primitives are defined it is possible to establish the vector geometrical representation hierarchy, as shown in the figure bellow.
![]() VECTOR REPRESENTATION - Hierarchy classes.
![]() Implementation UniverseWhile discussing the implementation universe it will be pointed out what data structures can be used to build up a Geoprocessing system. In this universe concrete programming decisions are treated with a large number of variations. These decisions can take into account the current application, the available algorithms to handle the geographical data and the hardware performance. For a deeper discussion about geographical operators implementation problems see Gutting et al. (1995). One of the main aspects to take into account in the implementation universe is the usage of spatial indexation structures. The access methods to spatial data are composed of data structures and recovery and search algorithms and they represent a determinant component in the system performance. A survey of this issues are presented at Cox Junior (1991) and Rezende (1992). These methods work over multidimensional keys and divide themselves according to the representation of associated data: points (e.g.: K-D trees), lines and polygons (e.g.: R trees and R+) and images (e.g.: quaternal trees). These and other methods has presented large performance improves in the geographical data access (mainly for the lines and points cases). The limiting factor for a large set of access methods studied is that they were designed to process in the main memory. In a large SGBDG it is required to access data efficiently in the secondary memory. This is true in both cases, for vector or raster data. For vector data Mediano, Casanova and Dreux (1994) presented a way to use an extension of B trees in order to show only the relevant geographical data for a certain scale, without running unnecessary procedures. This structure, called V-tree, allow to access multi resolution data and it is very convenient as support for skim methods in a large Database. ![]() Relationship among the universes in the modelThe “four universes modeling” paradigm (Gomes and Velho, 1995) considers that the mapping process among universes is not reversible and it admits alternatives. Following these relations will be discussed.
From the Real World to the Conceptual UniverseThe path from the real world to the conceptual universe considers some variations, as the application domain. In some cases, the mapping is direct, for instance, the satellite images and topographical or geophysical measures are naturally mapped to GEOCAMPO instances. In the city maps or political division map cases. The association with GEO-OBJECT classes and GEO-OBJECT MAPS is also direct. The thematic data collections can be used in two different interpretations, depending on its usage: when it is related to an inventory task (such as a vegetation map in the Amazon), it should be modeled as GEOFIELD instances (or, more specifically as a THEMATIC class). In detailed studies on averages and large scales (such as an economical-ecological) where each region is characterized by specific qualifiers, it is convenient that such a data collection is associated to a GEOOBJECT instance and GEO-OBJECT MAP. From the Conceptual Universe to its RepresentationThis mapping presents several nonexclusive options:
The literature has enforced the idea that a general GIS has to provide all the possible representation options. From Representation to ImplementationAs it was mentioned before, the implementation universe is a concrete programming decision. Following we will present some practical issues to be considered:
![]() SummaryTo better understand the relations among the different universes (levels) in the model, the table bellow presents several entity examples from the real world and their correspondence in the model.
![]() ![]() |