Abstract:

Because the complex global patterns could be structured from“bottomup”oriented selforganization,cellular automata(CA)have been increasingly applied in the simulation of urban growth and evolution.Spatial distances variables(driving forces)impacting urban growth were extracted from the multitemporal remotely sensed data,which were used to discover the transition rules of CA,with the auxiliary of field surveyed topographic maps and land use planning data.Based on the method of logistic regression,the logisticCA model was established.With a modeling package programmed in a GIS environment,the urban dynamics at the county and street block scale,Jiading District of Shanghai,were simulated from 1989 to 2006 with the logisticCA model.A pixel by pixel comparison method was used to produce a confusion matrix and the Kappa coefficient between the simulation results produced by the calibrated logisticCA model and the classified maps from remote sensing images.This study has demonstrated that the logisticCA model is particularly suitable to simulate urban growth with medium and small level areas.Finally,assuming that current urban development speed continues and no significant policy changes will happen in the future 4 years,the spatial pattern of Jiading District can be predicted by using the 17 years(1989~2006)historical data calibrated logisticCA model.