Coupling multi-agent model and GIS to simulate pine wood nematodedisease spread in ZheJiang province, China

摘要

A coupled method based on multi-agent model, remote sensing and GIS is described to simulate the forest diseasespread. The coupled model focuses on the temporal dynamics of the Bursaphelenchus xylophilus population at thelandscape scale. Each individual is modeled as an autonomous agent who behaves according to a set of rules includingspreading in the landscape, feeding on Pinus massoniana, sheltering in forest edges and dying, constrained by terrain,land cover and other variables. The model parameters are derived from remote sensing data and field measurements. Tenfactors, including damage degree of Pinus Massoniana, altitude and slope, are helped to build the transfer rules. Themain outputs are the dynamic disease distribution maps and survived pine population. Our method is applied andvalidated in DingHai distinct, Zhou Shan city of Zhejiang Province. Three Landsat TM images from the year 1991 to2006 are used for the pine information extraction. The extracted pine distribution map is used to compare with thesimulated surviving pine map. The results show that the coupled model can produce reasonable results and be used as avirtual experiment tool. However, it is difficult to simulate the human activities to help or prevent disease spread and thelong fly behavior of insect vectors. Therefore, there still exists some difference between the simulated results and the realdata. At the next step, those factors will be considered.