PIK Report No. 27
2. Common Integration Problems of the Projects
The common problems faced by any problem-oriented research project requiring integration can be classified by the overlying questions 'what has to be integrated?' and 'by what means can this integration be carried out?'. The most relevant questions identified by this method are collected in Tab.1 and discussed in the following.
ad 1.) Probably, this is not only the most difficult but also the most important integration problem. More than the other problems it represents a communication and a epistemological problem and there is definitely no unique answer and approach to this problem. The difficulties increase with the degree of interrelations between the disciplines, i.e. from the simple coupling of sectoral models where disciplines just represent boundary conditions for each other, up to the highly resolved processural dependence between differentiated processes traditionally analysed by single disciplines. The problem is an epistemological one as it involves disciplines with entirely different paradigms, e.g. critical rationalism vs. hermeneutics, and it is a problem of communication as the only way towards integration is by extended discussions and communications between disciplinary scientists. It is often stressed that disciplinary integration can only be achieved by the formulation of a common, if possible neutral, language which is acceptable for all partaking disciplines. Systems Analysis, Modelling and Simulation, representing the major approach at PIK, is claimed to be one such language.
# | Integration Problem | Need / Search for |
|---|---|---|
1 | Different disciplines/sectors | Common language |
2 | Different aggregation characteristics | Scaling and averaging techniques |
3 | Different levels of analysis | Macro variables, indicators |
4 | Uncertainties and disagreements | Boundaries, probabilities, risks |
5 | Normative and evaluation aspects | Critical levels, feasible sets |
6 | Interactivity for political decisions | Interface to/for policy makers |
ad 2.) The real world problem as discussed above involves a wide range of natural and anthropogenic processes which vary in their spatial, temporal, and functional aggregation characteristics. The range of interdisciplinary diversity from natural to socio-cultural processes is superimposed by ranges of intradisciplinary characteristics which often are analysed in terms of hierarchic discrete levels, e.g. leaf processes vs. biome responses to climate change or individuals vs. groups or institutions. This hierarchical approach - often called for by the structure of available data - generates scaling problems which - as a look at the examples just mentioned shows - are not easy to solve. Possible approaches include appropriate averaging techniques, homogenization, multiscale computations, and others. Similar to the problem of integrated disciplines, where the ever prevailing rationality has to be retained, the integration of scales has to take into account the specific spatial and temporal eigenscales of the processes involved.
ad 3.) One major task in any type of Systems Analysis is the formulation of macroscopic variables able to describe the problem and its interdependencies in an appropriate way. They may be considered as vehicles for solving the scaling and interdisciplinary problems mentioned above and as major representatives of the information structure of the entire model. Moreover the consideration of a variety of levels of analysis (e.g. heuristic description vs. theoretically well based principles) requires the usage of macroscopic variables able to connect these levels. The degree of resolution should be as fine as necessary and as rough as possible.
ad 4.) Integrated modelling is always faced with uncertainties within each individual subelement as well as with disagreements. As an example of the first, one can name the still open problems of cloud generating processes in the climate system whereas disagreements can be found mainly in the social sciences, e.g. the different theories for economic growth. Approaches include both external techniques such as sensitivity or uncertainty analyses of (deterministic) models and internalizing uncertainty through stochastic or fuzzy modelling techniques, construction of modelling toolboxes, or scenario building techniques by means of expert opinions. Thus, reference to events has to be replaced by reference to intervals and probabilities, the prediction of disasters by the risks of their occurrence.
ad 5.) It would be tautological to state that problem-oriented research should formulate problem solving strategies which extend from regional to global sustainability. As, however, it is not quite clear what actually can be considered to be a solution and as deciding this cannot be a task for science alone. Global Change research is always asked to include some evaluation techniques for orientation on scientifically determined critical levels. Care has to be taken to distinguish clearly between normative and ethical aspects on the one hand and functional and scientific questions on the other. Beside classical modelling approaches with a successive evaluation, new techniques like admissible set dynamics seem to be appropriate.
ad 6.) Finally, the problem solving aspect of Global Change research often requires the possibility to change the scenarios, parameters, critical levels, etc. in an interactive way. Therefore, pragmatic and efficient modelling and simulation techniques have to be used, which might even be used by policy makers interactively. For other purposes it might be more appropriate to carry out policy exercises which can help to develop scenarios or perspectives.
In the following we discuss the different projects in more detail where each project addresses a subset of the problems discussed above with different focuses and approaches. It should be stressed again that the different projects represent a fruitful variety which should result in a increasingly extending toolbox of integration techniques.
