Integrated assessment of agricultural and environmental
policies – towards a computerized framework for the EU
(SEAMLESS-IF)

Martin van Ittersuma, Frank Ewerta, Johanna Alkan Olssonb, Erling Andersenc, Irina Bezlepkinad,
Floor Brouwere, Marcello Donatellif, Guillermo Flichmang, Thomas Heckeleih, Lennart Olssonb,
Alfons Oude Lansinkd, Andrea Rizzolii, Tamme van der Walj, Jacques Weryk
a Plant Production Systems, Wageningen University, Wageningen, The Netherlands (seamless.office@wur.nl)
bLUCSUS, Univ. of Lund, Sweden; cFLD, KVL, Copenhagen, Denmark;dWageningen Univ., The Netherlands; eLEI,
Wageningen UR, The Hague, The Netherlands; fCRA-ISCI,Bologna, Italy; gIAMM-CIHEAM, Montpellier, France;
hUniv. of Bonn, Germany; iIDSIA-SUPSIA, Lugano, Switzerland; jAlterra, Wageningen UR, The Netherlands; kINRA,
Montpellier, France

Abstract: Agricultural systems continuously evolve and are forced to change as a result of a range of global
and local driving forces. Agricultural and environmental policies are increasingly designed to contribute to
the sustainability of agricultural systems and to enhance contributions of agricultural systems to sustainable
development at large. The effectiveness and efficiency of such polices in realizing desired contributions
could be greatly enhanced if it were possible to perform ex-ante assessments. The European Commission has
recently introduced impact assessment of its policies as an essential step in policy development. This paper
presents the design and first prototype of a computerized integrated framework to assess, ex-ante,
agricultural and environmental policies across a range of scales, from field-farm to region, EU25 and globe.
In this large integrated project 30 research groups work jointly on developing the SEAMLESS Integrated
Framework. Key requirements of the framework are that it is open, generic and transparent. This puts
stringent requirements on the software backbone of the project and on a modular set-up. The framework is
developed in close interaction with the targeted prime users, including the European Commission.

Keywords: integrated assessment, sustainable development, agricultural systems, software framework,
model components, ontology

1. INTRODUCTION
Agricultural systems around the globe
continuously change as a result of enlarging trade
blocks, globalisation and liberalization,
introduction of novel agro-technologies, changing
societal demands and climate change. Despite the
trend for liberalization, there is consensus in the
policy domain that policies are needed to support
sustainability within the agricultural sectors and
even more importantly to enhance the contribution
of agricultural systems to sustainable development
of societies at large. Agricultural, environmental
and rural development policies must contribute to
these aims, in a cost-effective and efficient
manner. Assessing the strengths and weaknesses
of new policies prior to their introduction, i.e. ex-
ante integrated assessment, is vital to target
policies at sustainable development. The European
Commission, for instance, has introduced Impact
Assessment of its policies as an essential step in
the development and introduction of new policies
since 2003 (EC, 2005). It explicitly calls for
assessment of the economic, environmental and
social impacts of policies and consultation with
stakeholders. By nature it implies a demand for
multi- and interdisciplinary research and tools,
which allow inclusion and evaluation of views of
different stakeholders.
Today actual use of research tools in the policy
domain for impact or integrated assessment is still
limited. Most of the approaches developed by
research which could potentially play a role are
still largely disciplinary and focused on specific
issues and hierarchical levels. Hence their use to
assess policies and innovations, which by
definition affect different hierarchical levels (e.g.
the globe, developing countries, EU25,
administrative region in a country, specific farms

and fields) and across economic, environmental
and social domains, is restricted. The gap between
analysis at micro level (farms) and macro level
(region or market) is still largely unresolved. Also,
as most of the research models are issue-specific,
possibilities for re-use in other assessments are
limited, whereas political agendas can evolve
rapidly. A further issue that limits integrated use
of different research models and tools is their ad-
hoc solutions in terms of software. Naturally this
has a strong impact on the possibility to re-use
existing research tools and to maintain them.
Lastly, end users and the way they will use the
tools are often not clearly identified and
determined.
Integrated assessment and modelling (IAM) has
been proposed by research as a means of
enhancing the management of complex
environmental systems. It is based on systems
thinking and a way to consider the different
aspects (biophysical, institutional, social and
economic) of a system under study (Harris, 2002;
Parker et al., 2002). IAM is an analytical approach
that seeks to gain insight from the analysis of
interactions. IAM has been defined as “an
interdisciplinary and participatory process
combining, interpreting and communicating
knowledge from diverse scientific disciplines to
allow a better understanding of complex
phenomena” (Rothman and Robinson, 1997).
The current paper introduces and presents the aims
of the SEAMLESS Integrated Framework
(SEAMLESS-IF) for an ex-ante, integrated
assessment of agricultural and environmental
policies and agro-technical innovations in the
EU25. SEAMLESS stands for System for
Environmental and Agricultural Modelling;
Linking European Science and Society. The main
features of its first prototype are presented.

2. SEAMLESS INTEGRATED
FRAMEWORK (SEAMLESS-IF)
2.1 Aims of SEAMLESS-IF
SEAMLESS aims at developing a computerized,
integrated framework (SEAMLESS-IF) to assess
and compare, ex-ante, alternative agricultural and
environmental policy options, allowing:
1. Analysis at the full range of hierarchical
levels (farm to EU and global), whilst
focusing on the most important issues at each
level;
2. Analysis of the environmental, economic and
social contributions of a multifunctional
agriculture towards sustainable rural
development.
3. Analysis of a broad range of issues and
drivers of change, such as climate change,
environmental policies, rural development
options, an enlarging EU, international
competition and effects on developing
countries.
SEAMLESS-IF will have the following specific
features and capabilities:
1. A multi-perspective set of economic, social
and environmental indicators of the
sustainability and multifunctionality of
systems, policies and innovations in
agriculture and agroforestry, derived through
so-called indicator frameworks facilitating
interactive and systematic selection of
indicators with users and stakeholders.
2. Quantitative models, tools and databases for
integrated evaluation of agricultural systems
at multiple scales and for varying time
horizons.
3. A software architecture, SeamFrame, that
allows reusability of indicators, models, data
and knowledge, also ensuring transparency of
models and developed procedures.
SEAMLESS-IF is applied and tested in two Test
Cases, one focusing on assessment of Common
Agricultural Policy reforms and trade
liberalisations as a consequence of WTO
negotiations, and a second on assessing local
implementations of environmental directives and
consequences of agro-technical innovations.
In short, SEAMLESS-IF aims to facilitate
translation of policy options into alternative
scenarios that can be assessed through a set of
indicators that capture the key economic,
environmental, social and institutional issues of
the questions at stake. The indicators in turn are
assessed using selected linkages of quantitative
models. These models have been designed to
simulate aspects of agricultural systems at specific
levels of organisation, i.e., point or field scale,
farm, region, EU and world. SEAMLESS aims at
an integrated use of these, partly, existing models.
SEAMLESS also assembles pan-European
databases for environmental, economic and social
issues. Some indicators, particularly social and
institutional ones, will be assessed directly from
data.
Linkage of models designed for different scales
and from biophysical and economic domains
requires software architecture, and a design and
technical implementation of models that allows
this. The software backbone of the project,
SeamFrame, serves that purpose.
Scientifically, the project aims at facilitating a
breakthrough in the integrated use of models,

enabling scaling. It targets at a modular approach
(‘mix-and-match’) and an open source software
architecture that allows use and re-use models,
databases and scenarios in the domain of
agricultural systems.
2.2 Users of SEAMLESS-IF
SEAMLESS-IF adopts a participatory
development trajectory. Within the project with its
30 partners and 150 researchers from a broad
range of disciplines this is done through working
with Prototypes and iterative cycles of
requirements analysis, testing and improving. In
parallel, Prime Users (Directorates General of the
European Commission) are involved in this
process through a User Forum. The User Forum
was established after an initial phase of bilateral
contacts and interviews, as well as small meetings.
In the design of SEAMLESS-IF six classes of
users are distinguished, i.e., coders, linkers,
runners, providers, viewers and players, with
distinct user requirements.

3. SeamFrame SOFTWARE
ARCHITECTURE
The main philosophy of SeamFrame is to create a
coherent simulation system by re-using a variety
of available, so-called ‘sources’, such as models,
databases, expert rules and analysis tools. All
sources need to implement a standard interface.
Figure 1 provides a schematic overview of the
overall design and its stacked component
architecture. SeamFrame is build upon the
OpenMI 1.0 standard (www.openmi.org). To
facilitate mixing these sources, SeamFrame
incorporates domain knowledge and semantic
meta-information. This allows for checking the
match between sources. This mix-and-match
philosophy is embedded in the SEAMLESS
OpenMI+ Framework Architecture (SOFA) which
forms the software spine of SeamFrame. It
facilitates both system development and system
evaluation. With SeamFrame, any ‘source’ that
implements the OpenMI interface can be linked. It
is envisaged that tools will be added for cross-
model debugging and sensitivity analysis covering
the whole chain of models.
SeamFrame will guide users choosing the
appropriate models, model/tool combinations and
data bases for the various policy evaluations of
agricultural systems. SeamFrame will provide a set
of pre-packaged applications for these decision
makers, and a graphical user interface. Users
belonging to the class coders, linkers or runners,
however, will use the modelling environment to
create and adjust models and data. Such scientific
tasks will be achieved using components from the
component toolbox: knowledge manager, a tool
for data and ontology manipulation, and the
modelling environment, for creating and editing
executable models. In Prototype 1 the overall
architecture of SeamFrame was designed, initially
implemented and populated with first essential
components and applications (see Section 5).
Semantic meta-information is organised in the
knowledge manager (Seam:KM) using the first
draft ontology for data and models (Athanasiadis
et al., 2006; Villa et. al., 2006). Also the design to
integrate modelling environments within
SeamFrame is made (allowing to ‘plug-in’
alternative modelling environments), with as
current choices MODCOM (Seam:MOD) for
biophysical models and GAMS (Seam:GAMS) for
farm economic and market models. The thus
created model-chain is deployed in the processing
environment. SeamFrame is developed using an
architecture centred software development process
allowing a staged delivery of pre-existing or new
and enhanced components (Van der Wal et al.,
2005).

4. INDICATOR FRAMEWORK
Baseline and policy scenarios are assessed and
compared in SEAMLESS-IF through their
characterisation by a set of indicators. These
indicators must capture the main features of
interest to users and stakeholders, about the
economic, environmental, social and institutional
issues at stake. So-called indicator frameworks are
developed to structure a broad range of indicators
and to facilitate their interactive selection by
stakeholders and users. Discriminating classes of
the initial indicator framework as listed in Table 1.
Indicators are assessed either through quantitative
model components (next section) or directly
Sources
SOFA
Modelling Environment Processing Environment
FSSIM -DM FSSIM-MP
Seam:GAMS
Seam:MOD
APES SEAMCAP
E
nd- user
A
pplications
Fram
ew
ork
A
pplications
S
eam
Fram
e
Seam:KM Seam:LINK Seam:PRES
E
nd- user
A
pplications
Fram
ew
ork
A
pplications
S
eam
Fram
e
Figure 1. SeamFrame architecture, its framework
applications and end-user applications

through data or semi-quantitative procedures (cf.
social and institutional indicators).
Table 1. Discriminating classes of initial indicator
frameworks. Within Prototype 1 SEAMLESS-IF uses
two classes, i.e., Dimensions of sustainability and
Levels of organisation.
Dimensions of sustainability
- biophysical, economic, social, institutional
Levels of organisation
- field, farm, region, EU25, globe
Domains
- sustainability, sustainable development
Themes
- goals, process of achievement, means
System’s properties
- viability, performance, capital, maintenance
5. MODEL COMPONENTS
5.1 Introduction
The first prototype of SEAMLESS-IF includes an
indicator calculator which draws information from
a model chain to compute selected indicators. The
model chain (Fig. 2) comprises the agricultural
sector model CAPRI that simulates supply-demand
relationships in the EU25 for agricultural
commodities; CAPRI derives information on
price-supply relationships from farm system
models (FSSIM) through an econometric
procedure (EXPAMOD). The farm models in turn
simulate farm behaviour and use agricultural
activities (e.g. crop rotations) assessed through a
mechanistic simulation model for agricultural
production and externalities (APES). Indicators
can be assessed through CAPRI, FSSIM and
APES, each at specific scales.
5.2 Agricultural Production and Externalities
Simulator (APES)
APES is a modular simulation model estimating
the biophysical processes of agricultural
production systems, at point level, in response to
weather and different options of agro-technical
management (cf. Van Ittersum and Donatelli,
2003). The processes are simulated in APES with
deterministic approaches mostly based on
mechanistic representations of biophysical
processes. This is done for a variety of regional
specific climatic conditions and soils. APES will
compute the yields, as well as several inputs and
externalities of crop rotations; both averages and
variability across years will be generated.
Currently first versions of weather (Donatelli et
al., 2006), crop, grassland, and soil water and
nitrogen components of APES have been
developed.
5.3 Farm System SIMulator (FSSIM)
FSSIM (Farm System Simulator) is a bio-
economic farm model developed to quantify the
integrated agricultural, environmental and socio-
economic aspects of farming systems. FSSIM
includes a data module, FSSIM-DM, which
computes the technical coefficients for agricultural
activities and FSSIM-MP, the mathematical
programming part of the model to capture
constraints and objectives (Deybe and Flichman,
1991; Janssen and Van Ittersum, 2006).
Applied at farm (micro) level, FSSIM seeks to
represent the actual farmers’ behaviour using the
knowledge of technical and socio-economic
constraints, the relation between production
factors, the amount of output obtained and the
costs of each production activity (= cultivation of a
crop rotation or livestock system in a specific
environment) and future market prices. This type
of models adopts a primal approach for describing
the technology, applying production functions,
partly derived from APES. FSSIM also uses
information from statistical databases and expert
knowledge for assessment of current activities. For
assessing alternative activities, the following
generators have been developed: Production
Enterprise Generator which generates alternative
crop rotations; Production Technique Generator
which generates alternative production techniques;
Technical Coefficient Generator which computes
the technical and economic coefficients for the
mathematical programming model. First versions
of these generators and a template for the
mathematical programming model of FSSIM have
been designed and implemented in SeamFrame.
Indicator Calculator
CAPRI
EXPAMOD
FSSIM-MP
FSSIM-DM APES
EU25
Figure 2. Models and model chain in Prototype 1 of
SEAMLESS-IF.

5.4 Agricultural Sector Model (CAPRI)
CAPRI (Common Agricultural Policy
Regionalised Impact) is an Agricultural Sector
model of the European Union (Heckelei and Britz,
2001). It is a comparative static equilibrium
model, solved by iterating supply and market
modules. The supply module consists of 250
regional non-linear programming models with a
highly differentiated set of activities allowing
direct implementation of most domestic policy
measures. Allocation is based on profit
maximising behaviour and calibrated multi-
product cost functions. CAPRI also estimates
regional nutrient balances and gas emissions with
global warming potential using a matrix of
coefficients linked with regional activity levels.
The CAPRI market module endogenously adjusts
EU- and international prices to achieve market
equilibrium. It also allows to asses the impact of
represent a large set of bi- and multilateral trade
policy instruments..
5.5 Econometric upscaling procedure for
micro-macro linkages (EXPAMOD)
The methodology envisaged to map the supply
behaviour of farm models (FSSIM) to the market
model (CAPRI) comprises the following sequence
of steps: 1) Simulation of FSSIM supply response
to price variations to obtain price-quantity data set;
2) Estimation of an econometric Meta-model
explaining supply response based on explanatory
variables, which determine FSSIM supply
response, but are also available for the whole EU.
3) Use of this Meta-model (EXPAMOD) to
extrapolate supply response to other farm types
and regions. 4) Aggregation of supply response to
level of CAPRI regions (administrative units). 5)
Calibration of regional supply modules in CAPRI
to aggregated supply response.
5.6 Anticipated model components to be
integrated in later prototypes
Next prototypes of SEAMLESS-IF are anticipated
to include the following components:
• Territorial models that enable assessment of
environmental and biodiversity indicators, as
well as visual quality of the landscape at
regional and lower levels.
• Rural employment model: an econometric
approach to assess effects of policies on
agricultural employment at EU25 and lower
levels.
• GTAP: a global trade (computerized general
equilibrium model) model, including global
database of production and trade, to analyse
the impacts of EU policies on the rest of the
world (Van Tongeren et al., 2001).
• Developing country models: computerized
general equilibrium model at national level
linked to farm household models (FSSIM) to
allow for assessing effects of EU policies on
agricultural production, environmental
impacts, poverty and rural development in
developing countries.

6. DATA AND TYPOLOGIES
In Prototype 1 first versions of pan-European
databases are available as ‘Sources’ (Fig. 1) for:
environmental data (soils, altitude and climate),
farming data and socio-economic data.
It is a major objective of the project to include data
that can be distributed freely. This means that it is
not always possible to include the original data
sets in SEAMLESS-IF due to property rights and,
in some cases, disclosure rules. For spatial data it
is therefore in some cases necessary to transform
the original data from vector data to grid data or to
a lower spatial resolution. For thematic data it is
likewise possible to distribute some data only in
aggregated format, for example by building
typologies of farms rather than distributing single
farm data.
Another major objective is to link the different
types of data, for example data from farm
statistics, data on the biophysical environment and
socio-economic data. This will be done by making
all data spatially explicitly linked to a 1x1 km2
grid, i.e. all data are available for that level,
although the spatial resolution for e.g. socio-
economic data is much less. A specific
achievement in relation to this is the development
of a statistical procedure to distribute the
information from farm economic statistics
available for administrative regions to biophysical
units with similar conditions for crop production.
Closely linked to this, and to enable upscaling
through statistical sampling procedures, typologies
are being developed for farming systems (based on
farm size, specialisation, intensity and land use;
Andersen et al., 2006), the bio-physical
environment (climate, soil, northing and oceanity)
and socio-economic characteristics (population
density, income, etc.). These typologies will also
provide a useful context to assess indicators
coming from the modelling results.
Datasets are currently characterised by metadata
(ISO-standards). Data are linked to models
through the use of the initially drafted ontology.
This enables future users to get interactively
insight in the quality of the stored data and to
identify appropriate use of the data. Additionally,

information will be included in the databases that
allows the user to explore the uncertainty related
to the data. This is crucial, especially for
aggregated data, to assess the robustness of the
final model results. Finally, tools and procedures
for updating the data will also be included in
SEAMLESS-IF to allow future users to
incorporate more recent information.

7. CONCLUSIONS
SEAMLESS targets at a working version of the
integrated assessment framework by 2008 for its
Prime users in the European Commission. At the
same time the software backbone of the project,
SeamFrame, is anticipated to provide an open
source means to facilitate integration of models
and other knowledge sources from different
domains and programmed in different
environments and languages. Finally, the different
components of SEAMLESS-IF are designed to
have standalone value. These components can be
used for targeted applications or serve as a starting
point for further scientific development. As such,
we aim that the integrated framework facilitates
condensation and synthesis of scientific
knowledge in the domain of agriculture and its
environment.

8. ACKNOWLEDGEMENTS
The work presented in this publication is funded
by the SEAMLESS integrated project, EU 6th
Framework Programme for Research
Technological Development and Demonstration,
Priority 1.1.6.3. Global Change and Ecosystems
(European Commission, DG Research, contract
no. 010036-2).

9. REFERENCES
Andersen, E., B. Elbersen, F. Godeschalk and D.
Verhoog, 2006. Farm management indicators
and farm typologies as a basis for
assessments in a changing policy
environment. Journal of Environmental
Management.
Athanasiadis, I. N., A.E. Rizzoli, M. Donatelli and
L. Carlini, 2006. Enriching software model
interfaces using ontology-based tools. 3rd
Biennial meeting of the Int. Environmental
Modelling and Software Society, July 9-12,
2006.
Deybe, D., G. Flichman, 1991. A regional
agricultural model using a plant growth
simulation program as activities generator.
Agricultural Systems, 37, 369-385.
Donatelli, M., G. Bellocci and L. Carlini, 2006.
Sharing knowledge via software components:
models on reference evatransporation.
European Journal of Agronomy, 24, 186-
192.
EC, 2005. Impact Assessment Guidelines,
SEC(2005)791, European Commission,
Brussels, 99 pp.
Harris, G., 2002. Integrated assessment and
modelling – science for sustainability. In: R.
Constanza and S.E. Joergensen (Eds.),
Understanding and Solving Environmental
Problems in the 21st Century, Elsevier, pp 5-
17.
Heckelei, T. and W. Britz, 2001. Concept and
explorative application of an EU-wide
regional agricultural sector model (CAPRI-
Projekt). Agricultural Sector Modelling and
Policy Information Systems. Proceedings of
the 65th EAAE Seminar, March 29-31, 2000
at Bonn University, Vauk Verlag Kiel,
Germany, Heckelei, T., H.P. Witzke & W.
Henrichsmeyer (Eds.), pp. 281-290.
Janssen, S. and M.K. Van Ittersum, 2006.
Assessing farmer behaviour as affected by
policy and technological innovations: bio-
economic farm models. Agricultural Systems.
Parker, P., R. Letcher and A.J. Jakeman, 2002.
Progress in integrated assessment and
modeling. Environmental Modelling and
Software, 17, 209-217.
Rothman, D.S., J.B. Robinson, 1997. Growing
pains: a conceptual framework for
considering integrated assessment.
Environmental Monitoring and Assessment,
46, 23-43.
Van der Wal, T., R. Knapen, M. Svensson, I.
Athanasiadis and A.E. Rizzoli, 2005. Trade-
offs in the design of cross-disciplinary
software systems. MODSIM Int. Conf. on
Modelling & Simulation, Melbourne, Dec.
2005.
Van Ittersum, M.K. and M. Donatelli, 2003.
Cropping System Models: Science, Software
and Applications. Special issue of European
Journal of Agronomy, 18, 187-393.
Van Tongeren, F., H. Van Meijl and Y. Surry,
2001. Global models applied to agricultural
and trade policies: a review and assessment.
Agricultural Economics, 26, 149-172.
Villa, F., M. Donatelli, A. Rizzoli, P. Krause and
F.K. Van Evert, F.K., 2006. Declarative
modelling for architecture independence and
data/model integration: A case study. 3rd
Biennial meeting of the Int. Environmental
Modelling and Software Society, July 9-12,
2006.