Sally Kleinfeldt (1), Jonathan Deckmyn (1), Pieter van der Wolf (2), Olav ten Bosch (2), and Claudio Paniconi (1)
1. CRS4, Via Nazario Sauro 10, 09123 Cagliari, Italy
2. Delft University of Technology,
Faculty of Electrical Engineering / DIMES,
The Netherlands
Environmental researchers have made several attempts to integrate desired components within (often proprietary) GIS. The difficulties of this approach --- particularly for applications involving 3 or 4 dimensions --- have been well-documented. Another approach has been to create in-house GIS which are more amenable to model integration requirements. This alternative, however, is not realistic for the average environmental research institute or department.
The purpose of this paper is to draw attention to the substantial literature that already exists on an analogous problem: that of integrating uncooperative, often proprietary tools in the world of computer aided design (CAD). In CAD, software infrastructures for building integrated environments, called CAD frameworks, have been developed. It is our feeling that environmental researchers could profitably adopt techniques that have been developed in the CAD field.
CAD framework techniques have been used extensively as the software ``glue'' that can turn a collection of individual tools into an effective and user-friendly integrated engineering environment. CAD frameworks provide a wealth of functions that not only support the construction of integrated environments, but also assist users in operating such environments [van der Wolf, 1994].
Historically, the first role allotted to CAD frameworks was that of common data repository, or design database. Data which is common to a number of CAD tools is stored only once in the repository, from where it can be used as input for all tools. This promotes tool interoperability. We emphasize that a design database merely provides a facility for storing and retrieving data; actual tool interoperability would require common formats for the data in order to avoid translation steps.
Subsequently, more functions were added to support the management of design data. CAD frameworks started to fulfill a second role, that of design data manager. A design data management system maintains information about the structure and status of the design to provide management support and enforce constraints on the design process. Graphical browse facilities were introduced to visually present this information to the user. In the CAD framework community this information about the design data is called meta design data, or simply metadata.
The third major role allotted to CAD frameworks was that of design process manager. With the increasing number of tools in today's CAD systems, there is a growing need to support the user in correctly executing these tools to perform tasks. On top of such a tool management service, the framework may provide design flow management, which assists the user in correctly and efficiently performing design activities according to a locally defined design procedure.
As a first example of applying CAD framework concepts to an integration problem in the environmental sciences, we will describe a system to provide data management and browsing services for a collaborative hydrologic research effort. We needed to create an environment within which all project data could be managed, browsed, and retrieved for further study. More importantly, we wanted the browsing feature to allow in depth examination of the available data using various visualization techniques, some from GIS tools and some from scientific visualization systems.
To address our data management needs,
we created an independent, centralized repository.
To address tool interoperability and
reduce the complexity associated with the heterogeneous nature
of the data, the repository data is stored
in standard formats (Freeform and HDF).
To address our browsing needs, we created a graphical user interface
(HYBROW, for Hydrologic Browser) [Kleinfeldt et al., 1996], shown in
Figure 1,
which supports visualization of the contents of
the repository.
We have encapsulated functionality from tools such as GRASS, AVS, Iris Explorer,
Gnuplot, and ftp.
HYBROW was
implemented in C and Tcl/Tk, a simple, public domain
scripting language with built-in facilities for creating X Windows
``widgets''.
A second example of applying CAD framework concepts is based on NELSIS, a design flow based CAD framework developed at the Delft University of Technology [ten Bosch, 1995]. The idea behind design flow management is to transform the informal idea of the structure of the design process as it resides in the user's mind into a formal description (a design flow) which is then used to further assist the user. In our case the design flow describes the basic steps in the modeling process, such as preparation and checking of inputs, model execution, and interpretation and analysis of outputs. The framework allows complex queries in an intuitive way. It doesn't impose any restrictions on data organization, nor on the use of graphical user interface tools.
Our first results using the NELSIS framework are described in Figure 2. The elaboration of a complete ``NELSIS-flow'' for a complex hydrologic model is in progress.
Kleinfeldt, S., J. Deckmyn, B. Cosyn and C. Paniconi, GIS and scientific visualization for hydrologic simulation. To appear in: Proceedings of HydroGIS '96, International Conference on Application of Geographic Information Systems in Hydrology and Water Resources Management, Vienna, April 16-19, 1996.
ten Bosch, O., Design Flow Management in CAD Frameworks. CIP-DATA Koninklijke Bibliotheek, Den Haag, 1995.
van der Wolf, P., CAD Frameworks: Principles and Architecture. Kluwer Academic Publishers, Boston/Dordrecht/London, 1994.