A methodology for enhanced flexibility of integrated assessment in agriculture

Agriculture is interrelated with the socio-economic and natural environment and faces increasingly the problem of managing its multiple functions in a sustainable way. Growing emphasis is on adequate policies that can support both agriculture and sustainable development. Integrated Assessment and Modelling (IAM) can provide insight into the potential impacts of policy changes. An increasing number of Integrated Assessment (IA) models are being developed, but these are mainly monolithic and are targeted to answer specific problems. Approaches that allow flexible IA for a range of issues and functions are scarce. Recently, a methodology for policy support in agriculture has been developed that attempts to overcome some of the limitations of earlier IA models. The proposed framework (SEAMLESS-IF) integrates relationships and processes across disciplines and scales and combines quantitative analysis with qualitative judgments and experiences. It builds on the concept of systems analysis and attempts to enable flexible coupling of models and tools. The present paper aims to describe progress in improving flexibility of IAM achieved with the methodology developed for SEAMLESS-IF. A brief literature review identifying limitations in the flexibility of IAM is followed by a description of the progress achieved with SEAMLESS-IF. Two example applications are used to illustrate relevant capabilities of SEAMLESS-IF. The examples refer to (i) the impacts on European agriculture of changes in world trade regulations and (ii) regional impacts of the EU Nitrates Directive in combination with agro-management changes. We show that improving the flexibility of IAM requires flexibility in model linking but also a generic set up of all IA steps. This includes problem and scenario definition, the selection and specification of indicators and the indicator framework, the structuring of the database, and the visualization of results. Very important is the flexibility to integrate, select and link models, data and indicators depending on the application. Technical coupling and reusability of model components is greatly improved through adequate software architecture (SEAMLESS-IF uses OpenMI). The use of ontology strongly supports conceptual consistency of model linkages. However, the scientific basis for linking models across disciplines and scales is still weak and requires specific attention in future research. We conclude that the proposed framework significantly advances flexibility in IAM and that it is a good basis to further improve integrated modelling for policy impact assessment in agricultur