基于数据同化的元胞自动机.pdfVIP

1007-4619 (2011) 03-475-17 Journal of Remote Sensing 遥感学报 The CA model based on data assimilation ZHANG Yihan, LI Xia, LIU Xiaoping, QIAO Jigang School of Geography and Planning, Sun Yat-sen University, Guangdong Guangzhou 510275, China Abstract: This paper presents a new method for calibrating urban cellular automata (CA) using ensemble Kalman fi lter (EnKF) of data assimilation. In CA modeling, the key issue is defi ning transition rules, which usually consist of many variables and pa- rameters. There are many uncertainties in determining parameter values and the transition rules are deterministic and unchanged during the modeling process. As a result, the model errors would accumulate continuously in the simulation. The paper introduc- es the ensemble Kalman fi lter of data assimilation into CA model, and a data assimilation CA model based on ensemble Kalman fi lter was established. After using the model, the paper can derive analyzed values by merging information from remotely sensed observations with CA model predictions, and modify the simulated results closer to actual situation based on the analysis values. The proposed model has been tested in Dongguan, a city in the Pearl River Delta of Southern China. Experiments indicate that the method can reduce the model error in the simulation and help to generate more reliable simulation results by comparing vari- ance, accuracy and kappa coeffi cient. Key words: cellular automata (CA), geography simulation, urban expansion, data assimilation, ensemble Kalman fi lter (EnKF) CLC number: TP79 Document code: A Citation format: Zhang Y H, Li X, Liu X P and Qiao J G. 2011