Using multi-agent modeling technique to regionalize key processes and patterns of sustainable agricultural cropping systems in the North China Plain

The North China Plain (NCP) covers an area of around 328,000 km2 and is one of the most important regions of cereal crop production in China. Wheat and maize rotations and one season cotton are the most common cropping systems. The region contributes at an amount of about 50% to the countries wheat production and about one third of maize yields. Crop production in the NCP was focused in the last decades on increasing yields to meet the growing food demand accompanied by the limitation of arable land as a result of urbanization rate i.e. of the Beijing District. Food production needs can nowadays only be achieved by the optimization of agricultural management, i.e. fertilizer input, irrigation, improved crop rotations. The focus on increasing yields raised serious environmental problems, like water shortage and pollution, air pollution and soil contamination. Hence the development of future land use system approaches improving these conditions is essentially. This may provide both a high production level as well as a protection of resources. The multidisciplinary collaborative International Research Training Group project (IRTG) “Modelling Material Flows and Production Systems for Sustainable Resource Use in Intensified Crop Production in the North China Plain”, funded by the Deutsche Forschungsgemeinschaft (DFG) and the Chinese Ministry of Education, was launched to detect the potential of adjustments in cropping systems and to further develop management practices for sustainable resource use and protection of environmental conditions while assuring a high yield level. The here presented research concentrates on the construction of a modelling framework of different spatial-temporal scales in order to regionalize the detected key features and the effects of changing land use patterns. In order to investigate our research objective, the regionalisation of key features towards sustainable agricultural and to improve productivity in the NCP (North China Plain) we primarily have to identify these core features. On the economic site we determine “farmer income” to be an appropriate factor. The ecological site is served by the determination of “water use efficiency” (WUE) and yield. Both factors have to be further verified on their plausibility for our research aim by project member discussion. Next an evaluation of multiple computational approaches towards their practicability was investigated. Relay on both existing GIS data in the projects AIES data base and additional data provided by our Chinese Agricultural University (CAU) colleagues. Primarily the Cellular Automata (CA) concept based on previous work as well as statistical analyses, Data mining and cell neighbourhood relations was investigated. This approach has to be denied due to insignificant neighbourhood relations and the fact that no appropriate computing environment was found. Secondly a model combination of Markov Chain and Cellular Automata as it is proposed by remote sensing software techniques (IDRISI 15.0) has been evaluated. This approach combines stochastic probabilities for cell transitions with classical GIS facilities. Again no statistic significance of spatial transitions due to farmer decisions is found. The 3rd actual approach Multi Agent System (MAS) is believed to be the most promising for several reasons. We chose NetLogo 4.0.3 as appropriate computing environment. First it includes GIS data extensions, above this it is a powerful free designable and Java programmable cross-platform user interface. The hypothesis is the following, Multi-Agents (farmers) acting in space, interfere and interact in spatial scales corrupting their entities (arable land) and thus their income. We have parameterised soil and defined agricultural activity zones. Agents now are aware of their own productivity value and compete with direct neighbours. For future purposes the ecological key features WUE and yield will be investigate by the use of the DSSAT crop model. As a group of scientist actually use the DSSAT crop model in varying plant sciences, certain expertises are generated in our project. This and the author personal expertise will help for a sudden parameterisation and integration of AEIS GIS dataset parameters.