【作者】 李爱民；

【导师】 李广云；

【作者基本信息】 解放军信息工程大学 ， 大地测量学与测量工程， 2009， 博士

【摘要】 随着经济的快速发展和城市化水平的提高,城市空间扩张与耕地资源减少之间的矛盾成为地学领域的研究热点。本文以城市建成区为研究对象,以卫星遥感影像为基础,系统研究了城市空间扩张与用地规模预测的理论与方法。重点探讨了基于遥感影像的城市建成区边界提取技术、建成区扩展分析的方法、元胞自动机的建成区扩展模拟方法、城市用地规模预测的理论与方法、城市建成区人口的获取方法。本文的主要工作和创新点包括:1.对本文的研究背景和研究意义进行了阐释。对城市建成区用地规模的确定方法进行了归纳总结。对基于遥感影像的城市建成区边界提取方法进行了总结和分析。2.分析了建成区在遥感影像中的特征,提出了基于遥感影像提取建成区边界的技术方案,以郑州市为例,进行了建成区边界提取实验,通过对比分析,得出近年来郑州建成区快速扩张的结论。3.通过对SPOT5多光谱影像的研究分析,提出了近红外、红色和中红外波段组合是建成区解译的最佳波段组合,为建成区边界的准确提取提供了理论依据。4.通过对非监督分类、监督分类、BP神经网络分类、建筑指数NDBI法、三指数合成法5种方法的实验对比,得出了在基于遥感影像的城市建成区解译过程中,三指数合成法是城市建成区边界提取的最佳方法的结论。如果原始卫星影像中没有中红外波段,宜采用监督分类法。5.系统总结了城市建成区扩展分析的理论与方法。数量分析有建成区面积变化量、面积增长率、城区扩展速率、扩展强度指标;形态分析有紧凑度指标、分维数指标、放射状指标、重心指标;空间差异分析有八个方向上建成区面积的平均值、方差和标准差,八个空间方位上不同时间段的扩展面积及扩展强度;扩展合理性分析有城区面积-城区人口弹性系数和异速生长模型。并从数量、形态、空间差异、合理性四个方面对1999?2007年郑州建成区空间扩展布局进行分析,得出一些建设性的结论,为未来郑州市土地总体规划提供了理论依据。6.基于数据挖掘思想,提出了基于新建建筑物的建成区扩展分析的方法,为城市扩展分析提供了新思路。7.根据城市建成区空间扩展的特点,结合国内外在城市CA领域的研究成果,提出了集成GIS的城市空间扩展CA模型的构建方案。重点包括CA要素定义、元胞时空数据库构建、系统框架及开发模式几部分。利用VB6.0与MapX5.0建立了与GIS无缝集成的城市扩展CA模拟系统,并利用系统对郑东新区的空间扩展作了模拟试验。研究表明,集成GIS的CA模型能够使模拟结果可视化,利用GIS能够对模拟结果进行空间分析。8.利用R/S分析方法对郑州市用地规模的发展趋势做了定性分析。研究得出,未来15年郑州建成区仍将保持继续增长的趋势。9.在学习传统方法的基础上,提出了一种确定Logistic模型中饱和值L的方法,即利用SPSS软件对Logistic模型进行L值的选优计算,以样本值与模型预测值的标准差变化是否趋于稳定作为饱和值选取的标准。并以郑州市区非农人口预测为例进行验证,实验表明,利用该方法建立的预测模型是可靠的。10.总结了基于单因素(时间、人口等)的用地规模预测方法。分别采用线性回归模型、复利模型、灰色系统的GM(1,1)模型、Logistic模型、二阶自回归模型、双曲线模型、异速生长模型(包括年末人口和非农人口两种测度)建立郑州建成区面积预测模型,预测了2008?2030年郑州建成区用地规模,对预测结果进行分析。分析得出,在建成区面积预测中,近期预测适宜采用标准差相对小的GM(1,1)模型,中期适宜采用异速生长模型,远期采用Logistic模型。11.提出了顾及多因素的城市建成区用地规模预测方法。BP神经网络法和多元回归分析法都是顾及多个因素的统计方法。利用郑州市1984?2005年统计数据建立了BP神经网络、多元回归、灰色GM(1,1)和Logistic四种预测模型,对2004?2007建成区进行了模拟预测。结果表明,顾及多个因素的方法预测精度较高,其中BP神经网络法优于多元回归分析法。12.针对传统城市人口统计方法的不足,在对城市人口遥感估算方法系统总结的基础上,提出了一种利用遥感和人口地理信息系统现代技术手段,获取城市建成区实体地域人口数据的方法。利用1984?2007年郑州建成区人口统计数据,采用6种预测模型预测了2008?2030年郑州年终人口和非农人口,对预测结果进行了分析,能够为郑州市人口控制和城市发展规划提供帮助。更多还原

【Abstract】 The inconsistency between urban space expansion and farmland resource decrease is a focusin the geoscience fields along with the fast development of economy and the exaltation of thelevel of cities. The dissertation makes a systemic research on correlative theory and technologyabout urban expanding and urban land use scale, taking urban built-up area as researchful objectand based on remote sensing image. It is studied as emphases of the technique of extractingurban built-up area’s boundary based on remote sensing image and the method of analyzingurban built-up area expanding and simulating built-up area expansion based on CA andpredicting urban land use scale and obtaining the population of built-up area based on remotesensing image. The main works and innovations are listed as follows:1. An explanation is given on the research background and significance of the dissertation.The methods of ascertaining urban land use scale are summed up. The methods of extracting theboundary of built-up area are summarized and analyzed.2. The characteristic of built-up area in remote sense image is analyzed. The project aboutextracting the boundary of built-up area is proposed. The boundary of Zhengzhou built-up area isextracted, and the conclusion of its land use scale expanding rapidly is made.3. The conclusion that the combination among the band 1 and 2 and 4 is best is proposedbased on analyzing SPOT5 multi-spectrum images. It provides academic base for extracting theboundary of built-up area truly.4. The conclusion that the method of three-index combination is best among the methods ofextracting the boundary of built-up area on the basis of the experiment on the methods what areunsupervised classification, supervised classification, BP neural network classification, NDBIand three-index combination. The method which is supervised classification is considered ifthere is no medium infrared band.5. The methods of analyzing urban bulit-up area expansion are summarized by the numbers.There are some indexes what are area change amount, area increase rate, extend velocity andextand intension on the aspect of quanity analyse, and some indexes what are immediate degree,fractal dimension, radiation shape and barycenter on the aspect of shape analyse, and someindexes what are average value, variance, standard error of built-up area area from eightdirection, expansion area and intension during differet time from eight space orientation on theaspect of space difference analyse, and some indexes what are flexibility coefficient between thecity zone area and population, allometric growth equation model on the aspect of rationalityanalyse. The situation of Zhengzhou space expanion during from 1999 to 2007 is analyzied, andsome useful conclusions are made. It provides the academic foundation for Zhengzhou toprogram urban land use.6. The new method of analyzing bulit-up area expansion on the basis of time when thebuildings are made up is proposed. It supplies a new thought for analyzing urban expanion.7. The project that making up the dynamic model of urban space expanion integratingCellular Automata with GIS is put forward based on some studies and the specialty of built-uparea space expanion. It includes three parts what are definition of CA factor, cell spatial-temporadatabase, system frame and programme model. Taken Zhengzhou as an example, aspatial-temporal dynamic model for urban growth by CA is designed, and the simulation system for urban growth of integrating CA with GIS by Visual Basic 6.0 and MapX 5.0 is developed. Itwas used to simulate urban growth in Zhengzhou East Area on the basis of remote sensingimages and land use maps. It can make the result of CA simulation visible and embedded analysebe made by GIS.8. Qualitative analysis about the development trend of Zhengzhou land use scale is madeby using R/S method. According to the result of the experiment, Zhengzhou land use scale willkeep increased trend in 15 years.9. A new kind of method is puts forward on the basis of studying the former. That isfounding corresponding Logistic model with different Parameter L by the software of SPSS, thenwhether the standard deviation between forecast value with fact value become steady or not isthe standard of ascertaining the Parameter L finally. Prediction of nonagricultural population incertain city is taken as an example. The experiment indicates the method is dependable.10. The methods of forecasting urban land use scale based on single factor are summarized.Those models, Linearity regression, compound interest, grey system, logistic, twain stepsself-regression, hyperbola, allometric growth model, are used to forecast urban land use scale.Eight prediction models about Zhengzhou bulit-up area are made, and the results of predictionare contrasted and analyzed. The conclusions those the grey system GM(1,1) model is used inthe near future ,the allometric growth model in the metaphase and logistic model at a specifiedfuture date are made.11. the method of predicting urban land use scale based on many factors is proposed. Boththe BP neural network prediction model and the multi-regression analysis model are two modelsBased on many factors. There are four models made up to predict urban land scale in Zhengzhoucity, what are the BP neural network prediction model,the multi-regression analysis model,GM(1,1)model and Logistic model. The result shows that the motheds based on many factorshave a high precision. It also shows that the BP neural network has better precision than themulti-regression model.12. After analyzing shortage of usual city vital statistics and summarizing themethods of city population estimation based on remote sense images, a new method ofobtaining population data in bulit-up area is proposed, which is based on remote senseimages and population GIS. There are six models made up to predict urban population inZhengzhou city in the future.更多还原