Dynamic simulation of land-use changes in a periurban agricultural system

Studying the driving forces of land-use change dynamics (LUCD) is important for understanding the change process. A spatially explicit simulation model helps to test hypotheses about landscape evolution under several scenarios. This paper presents a dynamic simulation model of land-use change (LUC) of the Nong Chok area in Central Thailand. Simulation of LUCD has been performed integrating remote sensing, geographic information systems (GIS), and the dynamic simulation toolkit. The model is a cellular automata model that has been developed on the basis of selected spatial and human driving forces. This study was conceived for the simulation of LUC, in particular, from paddy fields to fishponds. The model was run for 19 years from 1981 to 2000. Data describing present and historic land-use patterns were derived from aerial photographs. Transition functions were developed using the ID3 algorithm of the LUC data sets. The model uses as its input a land-use map (1981) and spatial and human variables: distance to canal and age, ownership, religion, education, and family size of the farmers. The results of the simulation showed substantial ability of the model to diffuse fishponds. To validate this spatial simulation model of LUCD, the simulated maps were compared with the reference land-use maps using a set of landscape indices: number of fishpond cells, patch density, mean patch size, edge density, fractal dimension, and mean nearest neighborhood. (Résumé d'auteur