OpenMI: Open modelling interface
J. B. Gregersen, P. J. A. Gijsbers and S. J. P. Westen
ABSTRACT
J. B. Gregersen (corresponding author)
DHI – Water and Environment,
Agern Alle 5, DK-2970, Hørsholm,
Denmark
Tel.: +45 4516 9200
Fax: +45 4516 9292
E-mail: gregersen@lictek.dk
P. J. A. Gijsbers
WL – Delft Hydraulics,
PO Box 177, 2600 MH, Delft,
The Netherlands
S. J. P. Westen
WSL – Wallingford Software Ltd,
Howbery Park, WallingfordOX10 8B,
UK
Management issues in many sectors of society demand integrated analysis that can be supported
by integrated modelling. Since all-inclusive modelling software is difficult to achieve, and possibly
even undesirable, integrated modelling requires the linkage of individual models or model
components that address specific domains. Emerging from the water sector, the OpenMI has
been developed with the purpose of being the glue that can link together model components
from various origins. The OpenMI provides a standardized interface to define, describe and
transfer data on a time basis between software components that run simultaneously, thus
supporting systems where feedback between the modelled processes is necessary in order to
achieve physically sound results. The OpenMI allows the linking of models with different spatial
and temporal representations: for example, linking river models and groundwater models, where
the river model typically uses a one-dimensional grid and a short timestep and the groundwater
model uses a two- or three-dimensional grid and a longer timestep. The OpenMI is designed to
accommodate the easy migration of existing modelling systems, since their re-implementation
may not be economically feasible due to the large investments that have been put into the
development and testing of these systems.
Key words | decision support systems, integrated catchment modelling, interface standard,
model linking, open source
INTRODUCTION
Managing environmental processes independently does not
always produce sensible decisions when the wider view is
taken. Therefore, it becomes important to be able to model
not only the individual catchment processes – such as
groundwater, river flow and irrigation – but also their
interactions.
Consequently, many existing hydrological decision
support systems use combined hydrological models as the
main building blocks. One of the earliest of these systems
was SHE, the European Hydrologic Model System (Abbott
et al. 1986), which supports integrated modelling of surface
water, unsaturated flow and groundwater flow. Since then,
numerous other similar systems have been developed,
where each system supports a fixed combination of specific
hydrological and hydraulic models. In many cases, these
systems are fulfilling the needs for integrated modelling.
However, in some cases, the limited number of available
combinations supported by the individual systems forces
the modellers to make undesirable compromises with
respect to creating an accurate representation of the
physical phenomenon that is being modelled. Naturally,
any system can be adapted to specific needs. Depending on
the underlying software architecture this may be either
difficult or fairly easy, but for most systems such tailoring
requires access to the source code of the hydrological
models involved. In practice, this means that such systems
are typically built by model providers using only a limited
suite of models for which source code is available. Even
doi: 10.2166/hydro.2007.023
175 Q IWA Publishing 2007 Journal of Hydroinformatics | 09.3 | 2007

when the model provider has access to a large suite of
models, the number of possible useful combinations
between these models means that in many cases a
combination requested for a particular project is not
available as an off-the-shelf product and it may not be
economically feasible to create such a system for a single
user or project.
The objective of the EU co-financed HarmonIT project
was to address these problems through the development of
an open modelling interface (the OpenMI) that will allow
the easy linking of existing and new models. The HarmonIT
project, led by CEH in Wallingford (UK), had 14 European
partners representing end-users, research institutes and
commercial model providers. Perhaps the most notable
fact is that the three commercial partners (DHI – Water
and Environment, WL – Delft Hydraulics and Wallingford
Software), who are all providers of some of the world’s most
widely used modelling systems and in their normal business
considered as competitors, were dedicated in sharing their
knowledge and contributed in close co-operation to the
development and promotion of the OpenMI standard.
Essentially, the OpenMI standard is a software com-
ponent interface definition for the computational core (the
engine) of the hydrological and hydraulic models. Model
components that comply with this standard can, without
any programming, be configured to exchange data
during computation (at run-time). This means that com-
bined systems can be created and can be based on OpenMI-
compliant models from different providers, thus enabling
the modeller to use those models that are best suited to a
particular project. The standard supports two-way
links where the involved models mutually depend on
calculational results from each other. Linked models may
run asynchronously with respect to timesteps and data
represented on different geometries (grids) can be
exchanged seamlessly.
The usefulness of the OpenMI standard relies on the
availability of compliant models. In other words, when the
number of relevant compliant models has reached a critical
level, it becomes attractive both to deliver new compliant
models and to create linked systems based on OpenMI-
compliant models. Consequently, one of the main require-
ments for the OpenMI architecture was that it should be
cost-effective to migrate models and that the architecture
should, at the same time, give freedom to model providers to
make their own optimal software designs. Most OpenMI-
compliant models will, for many years, be based on existing
models that are being migrated. Such models typically
consist of thousands of lines of Fortran, C or Pascal code
and re-programming is too expensive. Those models should
be able to run both in their normal environment and under
the OpenMI environment, without having to maintain two
different versions and without having to complicate the
calculation core with a great deal of OpenMI-specific code.
The selected approach to fulfil all these requirements was to
make a lean standard that is essentially an interface
definition, allowing developers to make their own design
choices. In order to separate the OpenMI-specific code
from the calculation code, a wrapper design pattern was
developed and a number of generic support libraries that
can speed up the migration process were developed.
The ambition for OpenMI is that the standard should be
accepted and used by a wide range of model developers and
model users and possibly become a new world standard for
model linking. Much effort was put into making compre-
hensive documentation and guidelines (Gijsbers et al. 2005;
Moore et al. 2005; Tindall et al. 2005) and the standard and
all tools and libraries were made available as open source
(SourceFORGE 2005). Undoubtedly, there will be requests
for improvement of the OpenMI standard when the larger
community starts using it. In order to meet such require-
ments an OpenMI association is currently being estab-
lished. This association will be open for everyone to join
and will be responsible for the maintenance and further
development of the standard.
In order to demonstrate the capabilities of the OpenMI
a complex system of integrated catchment modelling is
shown in Figure 1. Meteorological data from a number of
measurement stations are handled by a database system.
This system will provide precipitation and evaporation data
to rainfall–runoff models. The rivers are modelled by a
simple conceptual river model that will obtain inflow data
from the rainfall–runoff models. For a particular river
reach, a more detailed representation of the river flow is
required. This river reach is modelled by a physically based
hydrodynamic river model. The river model will obtain
inflow data for its upper boundary from the conceptual
river model and will provide inflow data for the conceptual
176 J. B. Gregersen et al. | OpenMI: Open Modelling Interface Journal of Hydroinformatics | 09.3 | 2007