The work performed in the course of this dissertation has been to define a systematic agricultural management response to environmental and economic conditions that is functional under hypothetical scenarios, especially involving climatic forecasts into the future. This was done through the use of the FARMACTOR/Expert-N coupled modelling framework that links agent-based management parameters with crop growth simulation, as the two are strongly interconnected. Starting with the completed FARMACTOR framework that had yet to be thoroughly tested, this work involved the verification of the modelling procedure, population of appropriate data resources for calibration and application, and the presentation of simulation experiments in peer-reviewed publication. The innovative linkage of agent-based management with biophysical simulation has led to FARMACTOR becoming a reference for international research on integrated economic/ecological study, impacting the scientific community through its unique contribution to analysis of anthropogenic landscape systems.FARMACTOR, as adapted in the course of this dissertation, has presented concepts that add to the robustness with which agroecosystem simulation is conducted on field and regional scale. Breaking away from the convention of static management input into crop models is an important step in this regard. Especially under scenarios of future climate change, dynamic field management lends to the plausibility of projected crop performance. If simulation modelling is to be an important tool in efforts to mitigate and/or adapt to climate change, elements such as dynamic management may be indispensable components of modelling frameworks. The impact of management has too great of an influence on agroecosystem functioning to be ignored.The effort in the course of this dissertation to systematically account for the likewise crucial factor of subspecies genetic variation is also an early example of improving agroecosystem simulation. As of the commencement of this work, agricultural species were, for the most part, simulated as just that, a species, when the variance of growth process within a species is a fundamental component of agronomy. Cultivar choice is one of the most important tools available to agricultural practitioners in terms of regional/localized agriculture. At least the simulation of multiple cultivars, or agricultural subspecies, is necessary to capture the heterogeneous responses to identical environmental conditions. This work has presented a sound methodology to account for breeding progress, based on observed trends in crop phenotypes, while also demonstrating a methodology for comparing results of the regional simulation of multiple cultivars.Spatial or temporal adaptation to climate is mandatory in terms of agricultural-sector profitability and food security, from local to global scales. Simulation modelling could eventually prove to be a useful tool in predicting the suitability of different crops or cultivars for unique biomes, whether in terms of agricultural intensification, producing more on a fixed land area, or expanding production into new areas. Simulation will most likely prove be an effective alternative to resource-intensive field trials, at the very least the two are complementary. This dissertation has, in part, demonstrated the potential for utilizing field experiments, to varying degrees of specificity, through model parameter optimization procedures, to produce local and regional projections of crop performance and adaptive measures likely to be undertaken by farmers.A statistical model developed alongside, and sharing the principals of environmental planting triggers incorporated in the agent-based model, was used to define a predictive model for maize planting dates throughout Germany. The two models achieved comparable accuracy, while differing in their advantages and drawbacks. The statistical model is not associated with a complete set of economic and biophysical attributes that can both be drivers of the bioeconomic model and informative outputs. Its advantage lies in its simplicity in regional applicability, able to predict (or project, if using future simulated weather), planting dates throughout the whole of Germany. The yield component of the statistical model demonstrates that the date of planting is a stronger driver of yields than the weather during the weeks that influence planting dates. Because maize is planted in spring, on bare fields, as opposed to wheat and other fall crops planted following the harvest of a previous crop, the statistical model is not as effective in predicting fall planting dates as FARMACTOR which can accurately simulate the harvest date of a crop preceding fall sowing.
Verknüpfung zu Publikationen oder weiteren Datensätzen
