Application of an integrated system dynamics and cellular automata model for urban growth assessment: A case study of Shanghai, China

Ma Yan

http://dx.doi.org/10.1016/j.landurbplan.2008.12.002
Ji Han, Yoshitsugu, Hayashi, Xin Cao
Landscape and Urban Planning Vol. 91, 2009, 133–141

Research Background
§By the accelerating urban growth, the accurate assessment of urban growth has become increasingly necessary.
§CA has been proved successful in assessing urban growth by viewing city as a complex system, however, CA model focus more on monitoring and simulating the spatial-temporal changes than on incorporating and interpreting the socioeconomic driving forces.
§Therefore, the integration of CA with other socioeconomic models is urgently needed.

Research Purposes
§To build a model which would be reasonable for evaluation of urban growth driven by human and natural factors at different scales.
§ To test the accuracy of the integrated model and specifies both socioeconomic variables and spatial variables.
§If the accuracy of integrated model is competent for urban growth simulation, then project the urban growth of Shanghai city up to 2020

Research Method
§In the study, an integrated system dynamics (SD) and CA model is developed with its framework. (showed in figure right)
§The problems are mainly focused in two aspects:
(1) A real change of urban land driven by socioeconomic variables
(2) spatial pattern of urban expansion affected by physical factors.

Research Results
§Accuracy test (right figure) showed that SD module is competent for urban growth simulation assessment.
§Kappa statistics between simulation results and actual land use map in 1989 and 2000 is 0.53 and 0.62, suggesting the parameters are reasonable and CA module is capable for simulating the spatial growth of urban land in Shanghai.

Thinking
§By setting the socioeconomic factors as simulation conditions, it is successful to improve urban growth simulation using CA model
§Besides the spatial simulation, further simulation could be done by creating agents based on CA simulation results
§SD model is possible to be combined with CA and agent-based system.