Environmental Modelling & Software 14 (1999) 493–494
www.elsevier.com/locate/envsoft
Editorial
Integration and re-use of Environmental Models
Andrea E. Rizzoli a, J. Richard Davis b
a IDSIA, Instituto Dalle Molle di Studi, Sull’Intelligenza Artificiate, Corso Elvezia 36, 6900 Lugano, Switzerland
b CSIRO, Land and Water, Canberra, Australia
This issue of Environmental Modelling and Software
contains a selection of papers on the advantages and
problems of reuse and integration of environmental mod-
els.
New knowledge arises from continuous evolution and
progress in all the fields of science. The field of environ-
mental modelling is no exception. Since the introduction
of computers as tools to support for modelling some dec-
ades ago, programmers and modellers have been looking
for ways to improve and reuse the millions of the lines
of code produced by previous researchers. Re-using
existing models is definitely a necessity. Most environ-
mental models are used only once, not because they are
inappropriate for other applications, but because of their
inaccessibility to other researchers and because of the
hurdles of integrating existing software into other new
applications. It has been more convenient to make a
fresh start in most new modelling projects.
Software technology has advanced considerably over
the last decade. Not only have object-oriented
approaches made software re-use easier, but the explos-
ively rapid development of the Internet has opened up
great opportunities for sharing models with a wider
range of potential users. The birth of many “on-line
model bases” (see ESS GmbH, 1998; Ghk and GSF,
1998 for example) is a response to the demand for ready
access to existing environmental models.
Model re-use is allied with model integration. Often
the purpose of model re-use is to develop an integrated
modelling system for decision making. In other words,
a Decision Support System (DSS). DSSs are an active
research area in Management Science (Dolk and Kotte-
mann, 1993), where various mathematical programming
models are integrated into a single DSS which help a
decision maker solve the different phases of a manage-
ment problem. These phases range from production plan-
ning to inventory control to resource allocation. The task
of an environmental manager is not that different. Com-
plex problems involving planning, diagnosis and
resource allocation are common. For example, a decision
about allocating river flows will have an impact on agri-
1364-8152/99/$ - see front matter. Ó 1999 Elsevier Science Ltd. All rights reserved.
PII: S1364- 81 52 (99)00 01 2- 2
cultural land-use management, on urban water supply
and on the health of aquatic ecosystems. Different mod-
els need to be integrated if the decision maker is to come
to a solution that accounts for all aspects of such prob-
lems in a transparent, reproducible manner.
In this issue, we have collected contributions which
are grouped around three issues:
1. the use of object-oriented technology in model inte-
gration;
2. the integration of models at different space and
time scales;
3. decision making supported by multiple models.
The papers by Maxwell, Neil et al. and Reed et al.
address the first issue. Maxwell presents a theoretical
framework, which describes the interfaces that are
needed if models are to interact in a “federation”. This
approach is useful if the modeller requires a convergence
between various model interoperability environments
such as the DoD HLA (DMSO, 1999), CORBA (OUG,
1999) and Microsoft’s COM (Microsoft COM techno-
logies Microsoft, 1999). The use of the COM method-
ology is presented in the paper by Neil et al. Here legacy
FORTRAN code has been encapsulated in an object-ori-
ented model of a dairy farm. Reed et al. present a
software engine which enables the modeller to abstract
data modelling from mathematical modelling, thus
improving model and data reusability.
The paper by Lau et al. deals with the second issue.
They describe a problem of natural resource manage-
ment in the face of high climate variability. Atmospheric
and terrestrial models work on different time and space
scales and integrating these models is clearly a problem.
The paper by Rebolij and Sturm spans both the first and
the second issues. They describe the integration of a
traffic model with an air quality model using software
component technology. They use a GIS for space scale
integration. A similar approach has been adopted by
Marquez and Smith who use a GIS module, a land use-
transport-environment module, and an air-quality mod-