Browsing by Subject "Modellintegration"

Now showing 1 - 1 of 1

Linking farm economics and hydrology: Model integration for watershed-level irrigation management applied to Chile
(2010) Arnold, Thorsten; Berger, Thomas
As largest user of fresh water, the agricultural sector must resolve conflict of objectives ranging from economic goals of farmers to societal and environmental targets. Research must deliver tools to manage these objectives simultaneously. Single disciplines have resolved numerous problems with disciplinary solutions. However, problems emerging from interactions and feedbacks between disciplines can only be assessed with interdisciplinary tools and managed by institutions that coordinate across departments. Such complex problems are becoming an epochal task for Natural Resource Management (NRM). A number of modeling tools exist for irrigation management at watershed level that quantify biophysical processes and water quality. Simultaneously, agricultural economics developed production planning methods for allocating water resources optimally. However, integrated planning support tools are not available that take into account both domains and their interactions. Within a larger research project, it was the objective of this Ph.D. project to develop and test methods that integrate two complex modelling softwares for irrigation management. The distributed runoff model WaSiM-ETH quantifies water flows and evapotranspiration. The dynamic land use model MP-MAS is a multi-agent system in which farmers use economic reasoning to derive cropping decisions under given environmental conditions. Furthermore, the MP-MAS software contains the bucket model EDIC, which parameterizes the distribution of water from rivers to individual farmers through the canal system. Finally, the MP-MAS software was extended with a crop yield model with complementary irrigation. Model integration is understood as service provided within a research context. This context is defined by the study region, the project setting and by the strategic decisions within the research project - such as the choice of partner institutions and disciplines. Within the Maule River watershed in Chile (Linares Province, Region VII), the project ?Integrating Governance and Modeling? assessed the use of water in agriculture. Empirical research questions as well as modeling software were also part of this research context. Integration requires the conceptual, the technical and the procedural level. Conceptual integration describes processes and interactions between farmers, the canal system as distribution infrastructure and the natural system. It also describes how farmers plan and produce within this environment. Here, scale-dependent processes like irrigation efficiency or access to water by individuals were scrutinized. Technical integration is the implementation of the conceptual system into source code, e.g. by adapting legacy software, and by creating a software layer for hierarchical coupling of all software components. Procedural integration is the calibration, analysis, error eradication and validation of these models within the research context. Calibration and analysis of integrated model components is a step-by-step procedure. For all relevant processes and interactions, empirical data was first compiled and cross-evaluated. Then, standalone model components were calibrated so that interactions were parameterized as boundary conditions that are consistent across all disciplines. Empirical data pinpointed conceptual inconsistencies in the description of interactions, and standalone models were improved together with project partners. Ultimately, model components were coupled in such ways that interactions can be analyzed dynamically at minimum model- and software complexity. The calibration process along transdisciplinary cause-effect-chains resulted in the improvement of disciplinary models and model results. For example, the relevance of access to water beyond legalized water rights became apparent when empirical data and models were combined. Also, the calibration of the EDIC model required consistent use of data from all four disciplines and improved the calibration of the MP-MAS model. For the WaSiM-ETH model, an irrigation module was developed that is consistent across scales and reflects the needs of extension workers. Finally, model integration and coupling is discussed as research process. The process of calibrating a model with four components is not only a technical challenge for modellers and data management, but also a procedural challenge with regards to cooperation beyond disciplinary institutions and cultures. The structure of the integration process should be robust against errors and equally facilitate knowledge transfer between disciplines, iterative calibration across disciplines. Success factors are suggested to reduce transaction cost during the integration process.