Uni-Bayreuth grafik-uni-bayreuth



Agent-based Model of Land Use Change in Soyang Watershed

TERRECO Cluster S-01

From 04/2012 to 03/2017

Principal Investigator: Thomas Koellner, Björn Reineking, Soo Jin Park, John Tenhunen
Staff: Ilkwon Kim

Abstract 2011: In the case of Soyang River basin, water quality has deteriorated due to land use change within the watershed. This study aims to extract variables which influence land use change, and predict future land use changes in the Soyang River basin using 1995 and 2000 land use maps. The prediction map is validated with the map of actual land use for 2006 to test the applicability of the land use prediction model, using relative operating characteristic (ROC) and variations of the Kappa index of agreements (Kno, Klocation, Kstandard) for major land use classes (urban, forest, agriculture). The forest class has a high accuracy for both distribution ratio and spatial agreement, while urban and agriculture classes have a high accuracy for their distribution ratio and a low accuracy for spatial agreement. These results demonstrate that the model can predict overall distribution ratio without predicting exactly where land use changes occur, because urban and agricultural changes are more closely related to socio-economic factors than environmental factors used in this research.


Keywords: land use change, land use simulation, logistic regression, CA-Markov analysis, Soyang River Basin


project description in detail from proceedings of 2011 TERRECO Science Conference GAP


Abstract 2013: Many factors such as climate change, new regulations, or local development plans affect land-use decision-making processes in the Soyang River Basin. To support policy makers, it is necessary to understand potential impacts of policy instruments on land-users decision-making, and potential future land-use and land cover change. However, it is difficult to represent these factors in a model that reflects the complexity of decision-making processes of land-users. Agent-based modeling (ABM) is one way to consider the diverse factors and emergences in agent decision-making processes. Simulation of land use and cover changes based on the ABM approach is needed to develop land use and land cover scenarios that are then evaluated in many dimensions in order to support future policy-making.

The aim of this study is to develop a multi-agent modeling approach that provides insight into responses of critically important agents determining land-use policies in the Soyang River Basin. Given the uncertainty and dynamic characteristics of land-use change due to human and environmental interactions, ABM is one means for simulating socio-economic and environmental impacts on land-use and cover changes in the region. For the initial step of building an ABM, it is necessary (i) to find patterns and factors of spatial and temporal land-use change in the past with spatial statistics and (ii) to establish characteristics of agents’ with local surveys and interviews. Multi-nominal logistic regression analyses illustrate the influence of social and environmental factors. Based on those inputs, the ABM is expected to simulate land-use and cover changes in the region under different socio-economic and environmental scenarios. Currently, a pilot agent-based model for land-use change is based on such results of statistical analyses. This model will be continuously modified to include additional processes of land-use change via stakeholder interviews that identify socio-economic drivers and major policy options for expected scenarios, and by local surveys to characterize and categorize agents and their decision making processes. From this research, expected scenarios on land-use and land cover change based on algorithms for decision-making processes of agents will be extracted. The results will aid in the evaluation of ecosystem services by providing input data in the form of land use maps to other TERRECO researchers examining scenarios with models.

Key words: Agent-based model, decision making, land-use changes, scenarios, NetLogo

Poster January 2013

last modified 2017-03-30