Results

Identifying and characterising different types of studies, depending on the aspect of CCIVA investigated by the study, was one important outcome of this work. The types of studies are: change observation, impact attribution, impact projection, vulnerability indication, adaptation options assessment, adaptation practice description, risk communication, risk perception assessment, and scenario development. Each type of study usually applied certain methods and tools. Across all sectors, the impact attributions and impact projections were by far the most often found types.

There are, however, sectoral differences. In the health sector, the impact attribution is most common and projections of impacts are rarely found, whereas  impact projections are dominant in the agriculture and fisheries sector as well as in the water resources sector. The studies furthermore showed that the kind of research questions addressed differed between the types of studies applied. For example, impact projection studies in the agriculture and fisheries sector were mostly concerned with crop production questions. In contrast, studies of type impact attribution mainly investigated the already observable impacts of climate change on coastal and marine ecosystems. Within the sectors, the different aspects of CCIVA addressed by the types of studies are rarely put together to present a comprehensive view.

Further knowledge gaps and research needs that had not been clearly presented in the IPCC Chapter are:

• Adaptation research: Only few of the studies actually investigated adaptation relevant issues, i.e., few were of type adaptation options assessment or adaptation practice description, or of type impact projection and actually took into account adaptation measures when projecting impacts. There was no study that considered costs of adaptation, institutional barriers to adaptation, or that considered synergies between adaptation and mitigation.
• Geographical focus and scale: Most of the studies mainly named Western European countries explicitly in the abstracts (especially the UK). We identified a set of mostly Eastern European countries that were not once explicitly mentioned. Most of the studies of a wider European context do not provide the spatial resolution of impacts that is necessary to inform local and regional adaptation.
• Uncertainty of projected impacts: We found that almost 1/3 of the studies of type impact projection using simulation models applied only one impact model to one climate scenario. These studies cannot make statements about the uncertainty of their simulated results with respect to the impact model used and climate scenario considered. In addition, the complexity of the models, scenarios and simulated results often make it often hard to compare results across studies. Of the remaining 2/3 of studies, most consider several climate scenarios and only few employ more than one impact model.
• Time scales considered: We differentiated three time periods of projections: a short term (until 2030), a medium term (2040-2060), and a long term (2070-2100). It showed that most studies of type impact projection provided results for the long term and only very few named impacts for several periods. Adaptation decision makers, however, need projections for the nearer future as well.
• Cross-sectoral and integrated research: Most studies (over 2/3) considered only climate change as a driver of change in their analyses. Drivers of change that were often considered besides climate change are socio-economic factors (including population change, technology, economic growth), land use, and atmospheric composition. There is thus a need for more cross-sectoral and integrated research.
• Other issues: No study considered mitigation scenarios (such as stabilisation scenarios) to assess the benefits (see IPCC AR4 WGII TS), nor possible sudden and larger changes in social-ecological systems. In addition, no study provided probabilistic results of possible impacts. 
  

One further interesting observation is that the concepts of vulnerability and adaptive capacity did not play an important role in the studies analysed. In fact, the knowledge on CCIVA generated by the analysed studies can be represented without these concepts. This finding is in line with the linguistic analysis of the different theoretical and operational definitions of these concepts also conducted within ADAM for the development of the general mathematical framework. Thereby it was found that there is little coherence between the theoretical definitions of vulnerability and related concepts and the methodologies that makes these concepts operational. From these two findings we conclude that CCIVA research should not be driven by theoretical definitions of these concepts, but rather by a clear definition of the subject matter and the research question analysed.