[an error occurred while processing this directive][an error occurred while processing this directive]
Spatial data, quantitative and qualitative models, and expert knowledge are now often included in computer-based spatial decision support systems (SDSSs) because users require a comprehensive system that supports complex decisions. SDSSs are developed based on predefined, multifarious logical decision trees that are focused on a specific question. SDSSs developed for environmental and social research, as well as other disciplines, must compare different scenarios.
A SDSS was developed for the GLOWA IMPETUS research program. This article describes the development and implementation of a specific GIS visualization tool, ArcGISDoubleMapPanel, that is part of the SDSS. The tool lets the user evaluate two multitemporal layers in one panel.
[GLOWA, or Global Change in the Hydrological Cycle, is a program to develop solutions for those who use and manage water resources in West African countries. The IMPETUS project, a joint venture of the universities of Cologne and Bonn, Germany, employs a holistic approach for analyzing and forecasting the effects of the global change on water resources in two river catchments in West Africa: the wadi Drâa in southeast Morocco and the Ouémé River in Benin.]
Choices and Considerations
For creating systems and interfaces for the GIS-based SDSS, Java was chosen as the programming language. Java's components and libraries acted as an interface between the GIS, remote sensing, decision support system (DSS), and geodatabase modules. Developing the source code in Java, these modules can be connected to create a comprehensive SDSS that answers specific questions.
Using this platform-independent programming environment, the Java developer can create user-friendly GUIs that allow decision makers to choose default expert input variables or customize SDSS parameters (e.g., model variables) based on their knowledge. Eclipse software development kit (SDK), which contains the Eclipse platform (Eclipse 3.2), was used as the project's programming environment because it comes with tools for Java programming and an environment for developing Eclipse plug-ins.
In addition to JavaBeans [the reusable software components of the Java 2 platform], specific GIS and remote-sensing components implemented in ArcGIS Engine were easily integrated into the developed source code. Geodata files in raster, vector, and alphanumeric formats were stored in file geodatabases. File geodatabases deliver high performance when handling large datasets and do not have an inherent size limit.
Using advanced SDSS techniques, design and development were accomplished using the Model-View-Controller (MVC) approach. [MVC is a programming design pattern that solves dependency problems associated with applications that contain a mixture of data access code, business logic code, and presentation code by separating core business model functionality from the presentation and control logic that uses this functionality.] Single modules of the SDSS are programmed as pipeline functionalities and processors and can be implemented into an integrated software framework.
Most decision makers who will use the system are subject area experts who have limited experience with computers generally, and GIS and SDSS software in particular. During the design process of computer-based decision systems, acquiring detailed information about the requests, knowledge, and personal needs of potential users was essential. Programming innovative SDSSs depends on assessing decision makers' knowledge of GIS, remote sensing, and models. The application needs to be easy to use. Nevertheless, the finished product should also benefit advanced users who have detailed knowledge of computer models and GIS and are able to implement additional functionalities into the decision-making processes. Therefore, the SDSS had to incorporate both standard and advanced GIS functionality. In addition to spatial analysis tools, SDSS interfaces should enable the importing and exporting of geodata and models as required.
The general architecture of the SDSS is illustrated in Figure 1. This SDSS development environment combines the state-of-the-art software and engineering knowledge with spatial data analysis. The expert system shell is placed in the center of the SDSS, which coordinates the whole system. It is responsible for information flows and directs control flows. The communication between the DBMS, Model Base Management System (MBMS), knowledge base, and spatial data handling unit is implemented via interfaces. The interface for a decision maker is usually a GUI that provides access to the system for spatial decision support.
While designing and developing question-specific SDSSs, Unified Modeling Language (UML) class diagrams can help describe the structure of the systems and be used as programming schedules for the developer. These class diagrams are static structure diagrams or flowcharts that show the classes of the system, their attributes, and their relationships.
After considering the requirements of potential users, several modules were implemented to supply functionality for the ArcGISDoubleMapPanel. In addition to common GIS tools that are integrated in the SDSS toolbar, the panel provides thematic and nonthematic raster and vector maps that can contain either two attributes and one time slider or one attribute and two time sliders (to compare two different time frames).
Figures 2 and 3 contain screen shots of the ArcGISDoubleMapPanel tool during preliminary development of the SDSS. Two PageLayoutBeans, which have toolbars with common GIS tools and time sliders (for visualizing multitemporal time steps), were used. By displaying two feature layers with different attributes that are thematically mapped, the user can intuitively compare these attributes. A Java slider (JSlider) component, located below the two maps, allows the user to view changes in the scenario over time. This slider has access to multitemporal feature layers and provides different time increments. Using a processor that accesses the specific time step of the layer, the two maps of the ArcGISDoubleMapPanel are updated (using refresh) during run time. Based on the classification applied to the map, layers change color in response to movement of the time slider. In Figure 3, the ArcGISDoubleMapPanel displays feature data classified by one attribute. The user visually compares two maps from different time spans. This provides an easy way to detect change over time. Each PageLayoutBean has its own time slider.
Modern computer-based spatial SDSSs are comprehensive and complex systems that compile individual decision steps in one overall software structure. They can be customized to the needs of individual users based on a given logical decision tree that was predefined by users and developers during the design process.
Using ArcGIS Engine, comprehensive ArcGIS functionalities and partly remote-sensing analysis were incorporated in the SDSS. ArcGISDoubleMapPanel can be used as a multitemporal scenario viewerfunctionality that is not yet available in most GIS software and SDSSs. This tool incorporates vector as well as raster data and provides ArcMap functionality for visualizing one or two attributes for different time periods based on a preselection. In addition to simultaneous visualization of multiple scenarios, the ArcGISDoubleMapPanel is also useful for performing change detection analyses, assessing current conditions, and predicting future situations.
About the Authors
Dr. Rainer Laudien is a postdoctoral research fellow in the Bundesministerium für Bildung und Forschung/Federal German Ministry of Education and Research (BMBF) project IMPETUS Westafrika at the University of Cologne. His major field of work is software development for spatial decision support systems by using GIS, model, and remote-sensing data.
Sebastian Brocks is a student assistant in the BMBF Project IMPETUS Westafrika at the University of Cologne. His major interest is software development, specifically integrating ArcGIS Engine components in custom applications.
Stefan Weyler is a student assistant in the BMBF Project IMPETUS Westafrika at the University of Cologne. His major interest is XML programming and system configuration.
Dr. Georg Bareth, a professor at the University of Cologne, heads the GIS and Remote Sensing Group in the Department of Geography. His major research interests are in mobile GIS applications and location-based services, GIS-based regionalization of N- and C-dynamics in ecosystems; analysis of multispectral, hyperspectral, and radar data for land-use mapping; the application of GIS and remote sensing for agricultural purposes; terrestrial laser scanning; and geographic databases.
Development and implementation of the SDSSs are carried out for the interdisciplinary research project IMPETUS (an integrated approach to the efficient management of scarce water resources in West Africa. See www.impetus.uni-koeln.de). The programmed systems are implemented in the Impetus Spatial Decision Support System (ISDSS), a Java/XML-based framework.
This study is part of that interdisciplinary scientific project and is supported by the BMBF under grant No. 01 LW 06001A and 01 LW 06001B as well as by the Ministry of Innovation, Science, Research and Technology of the federal state of Northrhine-Westfalia under grant No. 313-21200200.