Das Projekt "Assessment of the European Terrestrial Carbon Balance - Spatial interpolation of in-situ data by neural network approaches for the assessment of carbon stock and carbon stock changes in European forests" wird vom Umweltbundesamt gefördert und von Universität Hamburg, Arbeitsbereich für Weltforstwirtschaft und Institut für Weltforstwirtschaft des Friedrich-Löffler-Institut, Bundesforschungsinstitut für Tiergesundheit durchgeführt. Objective: CarboEurope will develop methods for the European Terrestrial Carbon Balance that allows spatial integration and analysis of geo-referenced data sets by applying neural networks and fuzzy techniques. For many decision processes and for causal inference statistical information has to be completed by spatially explicit information in mapped format. The main obstacle to do so is due to the heterogeneity of the individual data sets in terms of formats, spatial and temporal resolutions, regional coverage, statistical scales, or associated assessment errors. This ramification renders the integration of data and the simultaneous analysis difficult. Classical statistical tools such as multivariate analysis fail, as major assumptions and constraints are violated. Neural networks and fuzzy techniques add a new philosophy to causal inference, as they mimic human thinking and reasoning and allow for handling data associated with uncertainty. Those techniques are superior to traditional statistics as their set of constraints and assumptions is considerably low. Results: The integrative module of CarboEurope will result in maps showing a realistic and reliable spatial distribution of carbon stocks and stock changes within forested areas. The results of the pilot study will open the possibility to provide spatially explicit data sets for entire continental Europe. Contribution of University of Hamburg: - Literature research on fuzzy logic and neural networks, - Spatial analysis, - Programming algorithms for a prototype of neural networks.