本文提出一种以时空Logistic回归模型来预测城市扩展的新方法。其首先在传统Logistic回归模型中加入空间自相关结构构建空间Logistic回归模型,然后,利用漳州市区近20年(1989-2009年)的数据,建立不同时期城市扩展模拟的多个子空间Logistic回归模型Mi,再采用一次平滑指数法综合处理这些时间序列的Mi,构建出顾及空间复杂性和时间序复杂性的时空Logistic回归预测模型。新方法一方面克服了传统Logistic回归模型法受限于预测年份影响因素数据难以获取的缺点,另一方面由于模型考虑了城市扩展的长时间序列复杂性,即综合了城市扩展不同时期影响因素不同的情况,使它更接近城市扩展的实际,因而预测精度会提高。以福建省漳州市区为例,分别运用传统Logistic回归模型方法,在传统Logistic回归模型中单独加入空间自相关结构的空间Logistic回归模型法和基于时空Logistic回归模型的新方法这3种方法,对2009年城市扩展进行了预测分析。结果表明,基于时空Logistic 模型的新方法比传统Logistic回归模型法和空间Logistic回归模型法的预测精度都要好,总体预测精度分别为81.02%、83.82%和87.00%,预测城市用地的精度从63.59%提高到67.35%和73.34%,ROC曲线下的面积AUC从0.826提高到0.883和0.924。
We start this study aimed at building a new method of spatiotemporal logistic regression model to predict urban expansion. This method first established a space Logistic regression model by adding autocorrelation structure based on the traditional logistic regression model, then built the multiple sub-space Logistic regression model Mi of urban growth simulation of different stages by Zhangzhou City's nearly 20 years (from 1989 to 2009) data. After this work, a spatiotemporal logistic regression model which took into account the spatial complexity and temporal complexity was constructed by using single exponential smoothing to treat these time series sub-model synthetically. On one hand,this novel method has overcome the traditional shortcoming that the influencing factor data is difficult to obtain in the prediction year, on the other hand, the model considers the complexity of urban growth in the long time series, that is a combination of urban expansion in different periods of different factors situation, bring it closer to the actual urban expansion, which will improve the prediction accuracy. Zhangzhou City of Fujian Province was taken as an example in the study, and the urban expansion in 2009 was forecast by using three methods, i.e. traditional logistic regression model, space logistic regression model and spatiotemporal logistic regression model. The results showed that the prediction accuracy of the new method based on spatiotemporal logistic regression model was more better than others, for which the overall prediction accuracy were 81.02%, 83.82% and 87.0% respectively, and the sensitivity of urban land use prediction increased from 63.59% to 67.35% and 73.3%. Area AUC under the ROC curve raised in size from 0.826 to 0.883 and 0.924.
[1] 姚士谋,帅江平.城市用地与城市生长[M].合肥:中国科学技术大学出版社,1995.
[2] Lopez E, Bocco G, Mendoza M, Duhau E. Predicting Land Cover and Land Use Change in the Urban Fringe: A Case in Morelia City, Mexico[J]. Landscape and Urban Planning, 2001, 55 (4): 271-285.
[3] Li X and Yeh A G O. Neural-network-based Cellular Automata for Simulating Multiple Land Use Changes Using GIS[J]. International Journal of Geographical Information Science, 2002, 16(4): 323-343.
[4] Clarke K C, Gaydos L J, Hoppen S. A Self-modified Cellular Automaton Model of Historical Urbanization in the San Francisco Bay Area[J].Environment and Planning B, 1997, 24:247-261.
[5] Verburg P H,van Eck J R R,de Nijs T C M,et al. Determinants of Land-use Change Patterns in the Netherlands[J]. Environment and Planning B: Planning and Design, 2004,31:125-150.
[6] Verburg A, Soepboer P H, Veldkamp W, et al. Modeling the Spatial Dynamics of Regional Land Use: The CLUE-SModel[J]. Environmental Management, 2002,30(3):391-405.
[7] Bo Huang, Li Zhang, Bo Wu. Spatiotemporal Analysis of Rural-urban Land Conversion[J]. International Journal of Geographical Information Science, 2009,23(3):379-398.
[8] Wu F, Yeh A G. Changing Spatial Distribution and Determinants of Land Development in Chinese Cities in the Transition from a Centrally Planned Economy to a Socialist Market Economy: A Case Study of Guangzhou[J]. Urban Studies, 1997, 34(11):1851-1879.
[9] 吴楷钊,吴波. 基于空间相关的逻辑回归模型的城市扩展模拟[J]. 河南大学学报(自然科学版), 2010, 40(3):267-273.
[10] 张德南,张心艳. 指数平滑预测法中平滑系数的确定[J]. 大连铁道学院学报, 2004, 25(1):79-80.
[11] 王英安,岳天祥. 基于格网生成方法的山东省人口密度空间分布模拟[J]. 曲阜师范大学学报, 2004, 30(3),91:93.
[12] 王济川,郭志刚.Logistic回归模型方法与应用[M].北京:高等教育出版社,2001,33-55.
[13] 周小成,汪小钦,吴波,等.城镇扩张的多源遥感图像动态监测分析[J].地球信息科学,2008,10(3):332-337.
[14] Pontius R G Jr, Schneider L C.Land-cover Change Model Validation by an ROC Method for the Ipswich Watershed, Massachusetts, USA[J]. Agriculture, Ecosystems and Environment, 2001, 85,239-248.
[15] Lantz C A, Nebenzahl E. Behavior and Interpretation of Factors Affecting the Performance of Climatic Envelope Models[J]. Journal of Clinical Epidemiology, 1996, 49:431-434.
