DESIGNING GEOGRAPHIC ANALYSIS PROCESSES ON

THE BASIS OF THE CONCEPTUAL FRAMEWORK

GEOFRAME

Cláudio Ruschel, Cirano Iochpe, Luciana Vargas da Rocha

Universidade Federal do Rio Grande do Sul, Caixa Postal 15064,

91501-970 Porto Alegre, RS, Brazil

Jugurta Lisboa F.

Universidade Federal de Viçosa, Campus Universitário,

36570-000 Viçosa, MG, Brazil

Keywords: Geographic Information Systems, Geographic Analysis Process, UML

Abstract: The investment in geographic in

formation systems (GIS) is usually justified by their ability of supporting

the execution of geographic analysis processes (GP). The conceptual design of a GP makes it independent

of a specific GIS product and enables designers to define the process at a high level of abstraction using a

language that enforces a set of logical constraints and is yet easy to learn. On the other hand, in order to

support interoperability a GP conceptual model should be sufficiently generic to allow a GP definition to be

translated to any of the logical data models implemented by existing GIS commercial products. This paper

presents an extension to GeoFrame, a conceptual GIS framework that supports the conceptual design of

spatio-temporal, geographic databases (GDB). This extension is actually a conceptual GP data model

relying on a set of UML diagrams as well as on a methodology of how to apply them to analysis process

design. On the basis of the PGeoFrame-A, the definition of a GP starts by the identification of its associated

use cases. Both control and data flows are described by means of activity diagrams with the new modeling

constructs provided by UML 2.0. Input as well as output data introduced in the workflow definition are

described in detail through a class diagram. In this way, CASE tools based on UML can be adapted to

translate GP conceptual design to the specific scripts as well as macro definition languages of different

existing GIS products.

1 INTRODUCTION

Knowledge about the space we live in has always

been of a great value to mankind. A few centuries

ago the geographic information we counted with was

not accurate, scarcely arranged, and almost

unavailable. Nowadays, many of these limitations

are not present anymore. Geographic information

captured by techniques that guarantee high precision

can be found in abundance and represent almost all

regions of the earth in different projections as well

as scales.

Currently, spatial and descriptive information are

kept

ntegrated in geographic databases (GDB).

These information is usually presented as well as

processed by so-called geographic information

systems (GIS). The investment necessary to build a

GDB is usually justified mostly by the results that

can be achieved from the geographic analysis

processes (GP) to be carried out at system’s

production time. The complexity of a GP may range

from a simple query to an intricate algorithm of

spatial analysis.

A GP can also be understood as a workflow that

defi

nes a

partial order of execution of a set of GIS

operations. For instance, in the context of an

environmental application, a GP can rely on a set of

user-defined selection criteria to process the GDB

and suggest best locations for ecology-preserving

national parks. Depending on the GIS, the GP can be

programmed either directly at the user interface or

through a specific API that provides a library of

geographic operations.

Ruschel C., Iochpe C., Vargas da Rocha L. and Lisboa F. J. (2005).

DESIGNING GEOGRAPHIC ANALYSIS PROCESSES ON THE BASIS OF THE CONCEPTUAL FRAMEWORK GEOFRAME.

In Proceedings of the Seventh International Conference on Enterprise Information Systems, pages 91-97

DOI: 10.5220/0002533300910097

 SciTePress

The knowledge of the main GPs that will be

executed at system’s production time is of

fundamental importance to the design of the GDB as

well as by the selection of the GIS software and the

metadata configuration for the geographic datasets

that will be acquired. Therefore, GIS software tools

should support the conceptual design of GP.

Though, most products support neither end-user

interface nor API for GP conceptual design.

Usually, commercial GIS products support an

operation interface, at the logical level, that

implements a variation of the so-called cartographic

modeling technique as it was proposed by Tomlin

(Tomlin, 1991). In the cartographic modeling each

data set is considered a layer. The notation usually

represents layers by their names involved in a box,

while the functions (or operations) that act on these

data are indicated over oriented arcs between boxes.

The cartographic model of most products completely

hides the database sub-schema that lies below the

concept of a layer. At the GDB logical level, a layer

may be either a table, or a complex view created by

some join operation over a number of tables.

Several specific conceptual models for GDB

design have been proposed and improved during the

last years. In the literature, one can find the

GeoOOA (Kösters, 1997), GMOD (Oliveira, 1997),

MADS (Parent, 1999), and OMT-G (Davis, 1999)

among others. Most of them support the design of

static aspects of the GDB (e.g. classes and

associations). GMOD is one of the few models that

enable designers to represent dynamic aspects such

as the causal relationship between classes (e.g. the

occurrence of rain for a certain period of time may

cause an occurrence of a flood in a certain region).

On the basis of a so-called transformation diagram,

the OMT-G support the modeling of the internal

aspects of a process, including geographic analysis

operations, relying on its own semantics definition.

This paper presents a solution to the specification

of geographic analysis processes at the conceptual

level, regarding both external and internal aspects,

through an extension of resources offered by the

conceptual framework GeoFrame (Lisboa, 1999).

Since GeoFrame is based on UML, GPs can be

defined in a high level language that is independent

of a specific GIS product. By adapting UML CASE

tools, it is possible to support interoperability for a

GP by translating its UML definition onto a

definition accepted at the interface of any GIS that

supports the types of operations provided at the

conceptual level.

The remainder of this paper is organized as

follows. In section 2, the conceptual framework

GeoFrame introduced. In section 3 a classification of

geographic analysis operations is presented. Main

operation categories are supported by the GeoFrame

extension to GP conceptual design (PGeoFrame-A).

GP design with the proposed GeoFrame extension is

discussed in section 4. The steps of a GP design are

illustrated by means of an example in section 5.

Section 6 concludes the paper and points out to

future work.

2 THE CONCEPTUAL

FRAMEWORK GEOFRAME

GeoFrame is a conceptual framework with basis on

the formalism of object orientation that makes use of

the UML language. The framework concept adopted

in the GeoFrame is the one of a generic design in a

domain that can be adapted to specific applications

in order to serve as a pattern for construction of

applications.

The framework offers a class diagram, which is

specified in the PGeoFrame

package (shown in

Figure 1) (Lisboa, 2002), where each package has

the classes that are used as a basis for modeling of

classes of a GIS application. The data schema

produced with usage of this framework can be

denominated as UML-GeoFrame schema.

Field

Geographic

represent represent

0...1 1...*

PGeoFrame

NonGeographic

Object

Spatial

Complex

SpatialObj

Polygon

Line

Point

Isolines

Irregular

Points

Grid

OfCells

Adjacent

Polygons

GridOf

Points

TIN

Region

Geographic

description

continence

1...*

0...1 1...*

Field

Representation

Metadata

Phenomenon

Geographic

Geometadata

Object

Geographic

Figure 1: The GeoFrame’s class diagram

An extension to the PGeoFrame package, which

stands temporal aspects, was proposed by Rocha

(Rocha, 2001), who introduced the PGeoFrame-T

package, that imports the PGeoFrame. Therefore, the

GeoFrame user has two alternatives of modeling. To

model only the spatial aspects of a GDB, only the

PGeoFrame package must be used. Nevertheless, to

ICEIS 2005 - HUMAN-COMPUTER INTERACTION

express the temporal aspects as well, the

PGeoFrame-T must be used.

To specify the GDB conceptual schema of an

application, the user sets forth classes of application

as a specialization of the GeoFrame classes.

Afterwards he forms groups with more similar class

features in different themes, according to the

application requirements, whereby for each theme a

class diagram is developed using the UML

resources.

The objects that have a spatial component are

instances of the GeographicPhenomenon

class,

while the others, which are also denominated

descriptive objects, are instances of

NonGeographicObject classes. Following the sight

dichotomy principle of fields and objects introduced

by Goodchild (Goodchild, 1990), in the GeoFrame

the GeographicPhenomenon class is specialized in

the GeographicObject and GeographicField classes,

which are respectively represented by SpatialObject

and FieldRepresentation. According to what is

shown in the Figure 2, the spatial representation is

indicated through a set of stereotypes, introduced as

pictograms at the upper right corner of the rectangle

that indicates the class.

The conceptual framework GeoFrame differs

from the other conceptual models with regard to

types of geographical data when it searches a total

compatibility with the UML language. The UML-

GeoFrame schemes can be constructed with CASE

tools able to present pictogram-shaped stereotypes.

Figure 2: The GeoFrame’s stereotypes

So that the GeoFrame may also be used to

describe the dynamic aspects of a GDB it is

necessary to have an incorporation of resources, as

described below:

– To offer a catalog with geoprocessing

operations, in such a way that they may be used

in processes specification;

– Ability to express in a class diagram the

associations amid both original and derivative

classes that result from geographic analysis

processes;

– To offer a methodology that allows a

specification of the geographic analysis

processes, with usage of other UML resources

not explored yet, like the behaviour diagrams or

the processes expression in the classes diagram.

3 CLASSIFICATIONS OF

GEOGRAPHIC ANALYSIS

OPERATIONS

The spatial nature of geographical data allows that

geometric operations and topologic functions be

applied to them. The way such operations must be

arranged in groups is still a point at issue in the

geoinformation field. Due to a diversification of

concepts and nomenclatures on the geographic

analysis operations, the designer who wishes to

make clear the usage of these operations, while they

are still in the conceptual period, will be facing

knotty problems.

The basic set of operations has been formed from

the classification developed by Albrecht (Albrecht,

1996), whereby concepts developed by other authors

were also aggregated (Aronoff, 1989), (Câmara,

2000), (Chrisman, 1997), (Davis, 1999), (Open GIS

Consortium, 2001), (Tomlin, 1991) leading it to

suppression and addition of operations to that

classification. For every operation the possible entry

parameters and types of results have been

determined according to the GeoFrame expected

representations. We have opted to use

specializations for the data types, whenever possible,

so that when any field or object representation is

applicable, it is possible to make use of

GeographicObject and GeographicField classes.

Only if any spatial representation is applicable the

GeographicPhenomenon general class is used. The

following list presents a synthetic description of the

operations defined in the GeoFrame catalog

(Ruschel, 2003):

– Selection: or "Non-Spatial Selection", it restrains

the set of GeographicPhenomenon instances, on

which it is applied, for instances that fulfill the

selection attribute.

– Spatial Selection: it restrains the set of

GeographicObject instances, on which it is

applied, for the instances that fulfill a spatial

predicate related to a GeographicObject of

reference.

– Region Selection: similar to the Spatial

Selection, it uses a settled spatial predicate, the

“inside” topologic restraint, applicable to any

GeographicPhenomenon. The region is

determined by one and only Polygon instance or

by GeographicField.

– Classification or Algebra: it only handles with

those values associated to the

GeographicPhenomenon. This definition

includes the majority of Tomlin´s map algebra

operations (Tomlin, 1991).

DESIGNING GEOGRAPHIC ANALYSIS PROCESSES ON THE BASIS OF THE CONCEPTUAL FRAMEWORK

GEOFRAME

– Buffer: it establishes a region founded in the

distances related to a GeographicPhenomenon of

reference.

– Overlay is a boolean or mathematic operation

applied in a pair of GeographicPhenomenon.

When only GeographicObject instances are

involved, a geometric processing is carried out

and the result presents new instances including

the attribute set of the original instances.

– Voronoi Diagram: construction of a Tessellation,

according to a set of points, so that each polygon

may contain the points of the plan closer to a

specific place, instead of any other place.

– Slope: applicable from an instance of

GeographicField of continuous distribution, with

values that may be discretisized in plans.

– Viewshed: it classifies a GeographicField as

"visible" or "non-visible", according to a spot or

region defined at a determined elevation over the

ground.

– Spread: it is applied both to a net topologic

structure and a GeographicField instance. It

takes into consideration the existence of a

starting point and values, generically called a

cost, in its structure, having as a result paths of

minor cost.

– Transform: calculation of the coordinates

amounts of any GeographicPhenomenon for a

system of cartographic projection different from

the original.

– Distance: it returns the distance between two

GeographicObject instances.

– Centroid: attainment of a point in a secure way

within a polygon, useful for generation of

topology in vectorial GIS.

– Dissolve: when the spatial relationship of two

instances of a GeographicObject class is of

vicinity and both possess the same value for a

determined attribute, they are aggregated into

one only instance.

– Interpolation: according to Geographic

Phenomenon instances, an interpolation method

is applied (ex: polynomial regression, Fourier,

Kriging) to get another set of data as a result,

which inclusively may have another

representation format. Some methods may

require additional numeric parameters.

4 SPECIFICATIONS OF

GEOGRAPHIC ANALYSIS

PROCESSES

According to Booch (Booch, 1999), the UML

(Unified Modeling Language) is a language for

specification, mainly for complex systems of

software. However, it is also enough expressive to

model systems that are not software.

To model geographic analysis processes, we

have opted for the adaptation of RUP (Rational

Unified Process) methodology, for development of

software using the UML. The simplified method

described in (Quatrani, 1997) has been utilized.

Instead of starting the acquirement of necessary

data directly from the class diagram, RUP suggests

that such acquirement be started through the use

case diagram. Our attempt is to find out “what” the

system must do. In this diagram the actors, the use

cases and the relationships among the use cases are

identified.

This methodology will be presented in the

sequence together with an example in the basic

sanitation field. In our example, we attempt to

determine the water pressure surface in a supplying

system portioned with several reservoirs.

Generate Terrain Surface

Generate Piezometric Surface

Water Supply Department

Generate Pressure Surface

<<include>>

Figure 3: Initial use case diagram

The example presented in the Figure 3 shows an

use case diagram, where the use case “Generate

Pressure Surface”, requested by an user of the Water

Supply Department, includes the use cases

“Generate Terrain Surface” and “Generate

Piezometric Surface”. It means that the water

pressure depends on the height values of the natural

terrain and on the height of the piezometric line in

every spot.

At this point of modeling, preliminary activity

diagrams may be created to show the flow through

use cases, or inside a particular use case.

Contours

Surface

Generate Terrain

Reservoirs

Generate Piezo-

metric Surface

Systems

Calculate Surfaces

Difference

Water Pressure

Surface

Head Loss

Function

Figure 4: Example of preliminary activity diagram

ICEIS 2005 - HUMAN-COMPUTER INTERACTION

The example presented in the Figure 4 furnishes

a preliminary identification of objects to be used and

the actions to be taken. To generate the terrain

surface, a set of contours is required. To generate the

piezometric surface, one needs the water system

limits and the reservoir responsible of the water

supply for that system. A head loss function is also

required, obtained from the pipes properties. The

difference calculated of these two surfaces generates

the water pressure surface.

For the next phases of modeling the semantics

introduced in version 2.0 of UML (OMG, 2004) is

used. The modeling element Activity, specialization

of Class in the UML metamodel is defined as a

specification of parameterized behavior. Therefore,

an Activity may be represented in the class diagram

or have its behavior detailed in the activity diagram.

By applying this semantics to proposal of

GeoFrame extension, after it has been identified, a

Geographic Analysis Process (GP) must be modeled

as an UML Activity class and expressed in the class

diagram. This class possesses associations with

classes of the user model that furnish entrance

parameters and as a result of its instantiation, some

instances of geographic classes or not. In the

GeoFrame such kind of classes may be plainly

called Process.

The class diagram in the Figure 5 formalizes the

list of all elements that have been identified. The

objects that appeared in the preliminary activity

diagram now are arranged in classes. The GeoFrame

pictograms indicate the spatial representation of

each class, with exception of the pictogram that

shows a gear, thus indicating that the class belongs

to the Process type. So, the classes

WaterSupplySystem, Reservoir and Terrain supply

instances to the process CalculatePressureSurface.

As an exit of this process, new instances of the

WaterPressure class are created.

The details on behavior of the classes inserted as

Process in the user model appear through the

refinement of the activity diagram. At this level of

details the operations of geographic analysis should

already be evident. In the UML the behavior of an

Activity is characterized as a sequence of

subordinated units where each individual element is

an Action.

Terrain

Reservoir

System

Pressure

Water

WaterSupply

Calculate

<<instantiate>>

Surface

Pressure

Figure 5: Class diagram incorporating a class of

Process type

Just like a GP defined by the user, the GeoFrame

catalog operations should be modeled as activities

invoked in an activity diagram through an action of

the kind of "CallBehaviorAction". Taking into

consideration that such operations are already

implemented in the GIS software to be used for the

GP execution, there is no need to have them

detailed. Other actions, like the “Read/Write Action”

and “ComputationalAction” will be necessary to

complete the diagram. An Action is represented as a

round square in this diagram.

In the example of Figure 6 the operations Spatial

Selection, Buffer, Classification, Overlay and

Interpolation have been used.

CalcPressureSurface

Water Pressure Surface

Difference: Text

Linear: Text

GridOfCells

Interpolation

Overlay

GridOfCellsGridOfCells

elev: Variable

Classification

elev-[v]*3/100: Text

WriteVariable

ReadAttribute

elevation: Attribute

reference

Polygon

restriction

Buffer

contain: Text

SpatialSelection

Contour SetReservation Set

Systems Set

Figure 6: Activity diagram of a Process class

DESIGNING GEOGRAPHIC ANALYSIS PROCESSES ON THE BASIS OF THE CONCEPTUAL FRAMEWORK

GEOFRAME

Besides these operations, in this activity diagram it

is possible to highlight the application of the

following modeling elements (OMG, 2004):

– Object node: represented with an rectangle, it

holds all kinds of data that is involved in the

process.

– Data flow and pins: the pins (little squares) at the

extremities of the arches represent a temporary

object that flows between actions.

– Expansion region: it is represented as an activity

in a broken line and is executed as many times as

is the number of elements of an entrance

collection.

– Expansion node: it is situated at the edge of the

expansion region. The entrance node maintains

an element separated from the collection during

each execution of the region. The exit node

accepts one element of every execution of the

region, making available a collection in case of a

complete execution of the region.

To avoid any harm to GeoFrame framework

original structure, necessary adaptations to make

modeling possible have been developed in a new

package called PGeoFrame-A, which also imports

the PGeoFrame

package.

Metadata

Process

Phenomenon

Geographic

Geometadata

Object

Geographic

Field

Geographic

Object

Spatial

<<instantiate>>

represents represents

Field

Representation

0...1

parameter

1...*

0...11...*

PGeoFrame-A

NonGeographic

Object

Figure 7: Class diagram of the PGeoFrame-A package

Within this same logic, the designer who

attempts to represent analytical processes with the

GeoFrame (Rocha, 2001) must use the PGeoFrame-

A. Simultaneous utilization of processes and

temporality in the same model is the purpose of

future works.

The Process class has been incorporated to the

PGeoFrame-A

package, as shown in the Figure 8.

This class is associated with classes that provide

entrance parameters, the same way they can provoke

the creation of new instances of NonGeographicObject

and GeographicPhenomenon.

5 RELATED AND FUTURE

WORKS

A series of projects is under development in parallel,

like:

– GisCase; the GisCase project implements a

CASE tool, in free software, that allows

modeling of GP and, with use of components of

libraries that can execute geographic analysis

operations, code generation. When this code is

compiled and executed, it access a GDB where

the input data are read and the generated data are

stored. The initial implementation is being built

with the Poseidon for UML Community Edition

CASE tool (Gentleware, 2004) and the TerraLib

library (Câmara, 2004), generating code in XML

and C++. However, the architecture of this tool

supports implementation with other CASE tools

and libraries;

– InterSIG: the InterSIG project intends to

complement and extend the GeoFrame-T

framework to modeling of spatial-temporal GDB

and documentation of analysis patterns for GDB,

as a base for construction of a geographic

catalog system;

– ArgoCASEGeo: is also a CASE tool under

development which the main propose is to

explore how an Analysis Patterns Catalog can

improve the GIS users’ productivity and the

GDB quality (Lisboa, 2004). This tool

transforms conceptual data schemas into logical

implementation of ArcView GIS, Geomedia GIS

and TerraLib library.

Taken as possible future works related to

geographic analysis processes, one can mention

integration of PGeoFrame-A and PGeoFrame-T

packages; examination of applicability of other

resources introduced by version 2.0 of UML and not

utilized in this work; and the organization of

catalogs of GP, in a format analogous to the

GeoFrame analysis patterns.

6 CONCLUSIONS

This paper has presented a method to specify

geographic analysis processes compatible with the

UML language, so that the user can be able to

develop the model diagrams in a compatible CASE

ICEIS 2005 - HUMAN-COMPUTER INTERACTION

tool. With these tools it is possible to not only

construct the GBD schema but even generate

executable code.

This method is based upon a conceptual

framework, the GeoFrame, which offers resources

that simplifies the usage of UML in the elaboration

of GDB conceptual schemas. The proposed

extension follows the same line, except that now it

simplifies the usage of UML at the specification of

geographic analysis processes.

The solution presented as a whole has been

incorporated to the PGeoFrame-A package, which

imports the contents from the original framework,

the PGeoFrame package. In a synthetic way, the user

of this framework who may choose to use the

PGeoFrame-A will find semantics that supports the

expression of geographic analysis processes,

together with a catalog of expandable operations.

Besides that, with utilization of GeoFrame it is

possible to develop a catalog of analysis patterns

that will progress along the time it is used. In a

similar way, our intent is to offer together with the

GeoFrame extension for specification of GP a basic

catalog for operations. This catalog has to be

independent of software and be able to be enlarged

by the user, according to the requirements of the GP

that may be implemented on the GDB.

ACKNOWLEDGEMENTS

This work has been partially supported by CNPq

(Brazilian National Research Council), the Brazilian

governmental agency for scientific and technological

development.

REFERENCES

Albrecht, J., 1996. Universelle GIS Operations for

Environment Modeling. In Proceedings of the 3.

International Conference on Integrating GIS and

Environmental Modeling. Santa Barbara.

Aronoff, S., 1989. Geographic Information Systems: a

management perspective. WDL Publications, Ottawa.

Booch, G., Jacobson, Y., Rumbagh, J., 1999. The Unified

Modeling Language User Guide. Addison-Wesley,

New York.

Câmara, G., et al., 2000. Towards a Unified Framework

for Spatial Data Models. J. Braz. Comp. Soc. Porto

Alegre, 7(1), 17 – 25.

Câmara, G., Onsrud, H., Monteiro, A.M.V., 2004.

Eficcacious Sustainability of GIS Development within

a Low income Country: The Brazilian Experience.

INPE. In: www.dpi.inpe.br/terralib.

Chrisman, N., 1997. Exploring Geographic Information

Systems. John Wiley & Sons, New York.

Davis Jr, C. A., Laender, A. H. F., 1999. Multiple

representations in GIS: materialization through map

generalization, geometric, and spatial analysis

operations. In Proceedings. 7th ACM GIS, Kansas

City, 60--65.

Gentleware, 2004. Poseidon for UML.

www.gentleware.com.

Goodchild, M., 1990. Geographical Data Modeling. In A.

Frank, M. Goodchild, Two Perspective on

Geographical Data Modeling. NCGIA, Santa Barbara.

Kösters, G. et al., 1997. “GIS-Application Development

with GeoOOA”. Int. Jounnal of GIS, 11(4).

Lisboa Filho, J., Iochpe, C., Borges, K. A., 2002. Analysis

patterns for GIS data schema reuse on urban

management applications. In CLEI Electronic Journal,

v.5, n.2.

Lisboa Filho, J., Iochpe, C., 1999. Specifying analysis

patterns for geographic databases on the basis of a

conceptual framework. In Proceedings 7th ACM GIS,

Kansas City, 7-13.

Lisboa Filho, J., Sodré, V. F., Daltio, J. Rodrigues Jr, M.

F., Vilela, V., 2004. “A CASE tool for geographic

database design supporting analysis patterns”.

Conceptual Modeling for Advanced Application

Domains. Proc. of ER2004 Workshop on Conceptual

Modeling for Geographic Information Systems

(CoMoGIS), Shanghai, China. LNCS 3289.

Object Management Group, 2004. UML 2.0

SuperStructure Specification. www.omg.org

Oliveira, J.L., Pires, F., Medeiros, C.B., 1997. An

Environment for Modeling and Design of Geographic

Applications. GeoInformatica, v.1, Kluwer, Boston,

29-58.

Open GIS Consortium., 2001. The OpenGIS abstract

specification, topic 1: feature geometry, version 5..

www.opengis.org.

Parent, C. et al. “Spatio-temporal conceptual models: data

structures + space + time”. In Proc.7th ACM GIS,

Kansas City, 1999.

Quatrani, T., 19997. Visual Modeling with Rational Rose

and UML. Addison-Wesley.

Rocha, L. V., Edelweiss, N., Iochpe, C., 2001. GeoFrame-

T: A Temporal Conceptual Framework for Data

Modeling. In Proc. 9th ACM GIS, Atlanta, 124-129.

Ruschel, C., 2003. Extending the Framework GeoFrame

for supporting Geographic Analysis Processes. Porto

Alegre: PPGC-UFRGS. Master Degree Dissertation

(in portuguese). www.inf.ufrgs.br/~ciochpe.

Tomlin, C. D., 1991. Cartographic Modeling. In: D.

Maguire et. al. Geographical Information Systems.

Longman, 362--374.

DESIGNING GEOGRAPHIC ANALYSIS PROCESSES ON THE BASIS OF THE CONCEPTUAL FRAMEWORK

GEOFRAME