基于元胞自动机民勤绿洲湖区荒漠化演化预测

民勤湖区是民勤绿洲中生态环境最为恶劣的地区,土地荒漠化问题十分突出。以民勤绿洲湖区为例,解译1992年、1998年、2002年和2006年TM卫星影像,分析其荒漠化动态变化情况,利用ArcObjects模块结合地理元胞自动机理论构造荒漠化动态模拟模型,通过对比2006年的预测数据与实际数据,对模型进行参数调整和预测检验。预测结果表明,模型预测的准确性达到90%。最后对2012年该区土地利用状况做出预测,进而对荒漠化的发展趋势进行预测分析。     

城市扩展元胞自动机多结构卷积神经网络模型

传统的城市扩展元胞自动机(CA)模型是基于单个元胞的变量信息挖掘来构建转换规则的。针对这一问题,本文基于多结构卷积神经网络提出从区域特征出发且顾及区域多尺度特征挖掘转换规则的城市扩展元胞自动机模型(MSCNN-CA),并以武汉主城区和上海浦东新区为例,模拟了两个试验区2005—2015年期间城市扩展过程。模型验证表明:与逻辑回归和神经网络相比,本文构建的3个单一结构的卷积神经网络元胞自动机(CNN-CA)模型在4个指标(Kappa系数、FoM(figure of merit)值、命中率(h)和错误率(m))上都有不同程度的提高。特别是FoM指数,在武汉主城区提高了23.3%～29.4%,在上海浦东新区提高了20.3%～28.5%。此外,MSCNN-CA模型与3个单一结构的CNN-CA模型相比,在各个指标上也有所改善,FoM指数在武汉主城区提高了0.8%～4.8%,上海浦东新区提高了2.8%～7.8%。两个试验区的模拟结果表明:相比传统CA模型,基于多结构卷积神经网络的城市扩展元胞自动机模型(MSCNN-CA)能够有效提高城市扩展模拟的精度,更真实地反映城市扩展空间演变过程。相比单结构的...     

基于多智能体的城市人口分布动态模拟与预测

城市人口分布变化过程是复杂的动态系统,掌握其规律在城市规划和社会可持续发展中有重要意义。用相互作用的多智能体系统(MAS)、元胞自动机(CA)环境及城市人口密度模型构建精确到街道的城市人口分布模型,并以长沙为例,分析城市人口分布的演变过程,为相关的调控提供决策依据。研究结果表明,其模拟的城市人口分布格局与实际情况吻合较好,在多种因素的影响下,长沙市将形成“市中心人口快速增长,近郊区人口缓慢增长,沿湘江畔、沿五一大道及岳麓区高新技术开发区人口密集”的发展格局。与以往的模型进行相比,所获得的模拟结果精度更高,更接近于实际的空间分布格局。     

顾及轨道交通影响的浙中城市群土地利用多情景模拟与分析

土地利用变化受到地形地貌、自然环境、城市规划和经济发展等的影响,预测其未来情景对政策调整具有重要的参考意义。元胞自动机模型是模拟和预测不同规划政策下土地利用变化的常用方法。本文基于GlobeLand30数据集,利用浙中城市群2000-2010年土地利用变化校准FLUS模型,并模拟2010年土地利用格局,其总体精度、Kappa系数和图形优化(FOM)分别为89.74%、82.69%和29.86%。采用马尔可夫链预测2030年各类型土地总量,利用FLUS预测一般条件下(常规情景)和城市轨道交通规划站点影响下(轨交情景)浙中城市群未来土地格局。结果表明,在5 km范围内城市轨道交通站点对建设用地增长影响较大,在该区域轨交情景比常规情景面积增加45.25 km^2、且主要发生在城市边缘区。建设用地扩张主要通过侵占优质农田实现,轨交情景5 km范围内农田转化为建设用地比常规情景增加33.34 km^2,建设用地扩张强度高于常规情景,其中最低扩张强度以上占比高于常规情景3.70%。景观指数表明,2种情景中林地、草地和水域格局具有较高相似性。本研究表明,综合使用FLUS、遥感、GIS等技术方法,能够准确模拟和预测不同规划条件下未来土地利用格局,并为规划和政策调整提供高可信空间数据。     

Blood metal levels and metal-influenced immune functions of harbour seals in captivity

Immunological blood parameters and the effects of environmental pollutants on the immune system are important to assess the health status of seals. Animals living permanently in seal centres are useful for development and validation of diagnostic tools for free-ranging animals. In this study, parameters of cellular immunity as well as metal concentrations in blood and metal influence on cell proliferation of seven seals from a seal centre were investigated repeatedly using multi-element analysis and a lymphocyte proliferation assay. The metal concentrations, except for tin and chromium, were in general comparable to those of free-ranging animals of the North Sea. The unstimulated and mitogen-stimulated lymphocyte proliferation showed strong intra- and inter-individual variability, which reflected variability in activation of the immune status. Furthermore, both immunosuppressive and stimulative influences of metals on lymphocytes were found. Summarising, the methods used in this investigation provided useful information on these animals, and their application to free-ranging animals can be recommended.     

基于最大发展潜力值的元胞自动机城市扩张模型

在城镇化发展水平及土地开发评价的基础上,评价待扩张区域土地转变潜力分值,基于最大转变潜力分值,提出一种有别于传统元胞自动机模拟城市扩张的新方法。该方法与传统元胞自动机模拟城市扩张相比在迭代方式上存在不同,它克服了传统元胞自动机因阈值设置不同而导致结果不确定的难题。分别将该方法与传统元胞自动机模型应用于武汉主城区,模拟了其2003年到2013年的城市扩张情况,最后,将模拟结果与实际土地利用现状图进行对比发现改进后的新方法在模拟精度上大大提高。     

基于神经网络的单元自动机CA及真实和忧化的城市模拟

提出了一种基于神经网络的单元自动机（ＣＡ）。ＣＡ已被越来越多地应用在城市及其它地理现象的模拟中。ＣＡ模拟所碰到的最大问题是如何确定模型的结构和参数。模拟真实的城市涉及到使用许多空间变量和参数。当模型较复杂时，很难确定模型的参数值。本模型的结构较简单，模型的参数能通过对神经网络的训练来自动获取。分析表明，所提出的方法能获得更高的模拟精度，并能大大缩短寻找参数所需要的时间。通过筛选训练数据，本模型还可以进行优化的城市模拟，为城市规划提供参考依据。     

Effects of crab–halophytic plant interactions on creek growth in a S.W. Atlantic salt marsh: A Cellular Automata model

The Bahía Blanca Estuary (38° 50′ S, and 62° 30′ W) presents salt marshes where interactions between the local flora (Sarcocornia perennis) and fauna (Chasmagnathus granulatus) generate some kind of salt pans that alter the normal water circulation and condition its flow and course towards tidal creeks. The crab–vegetation dynamics in the salt marsh presents variations that cannot be quantified in a reasonable period of time. The interaction between S. perennis plant and C. granulatus crab is based on simple laws, but its result is a complex biological mechanism that causes an erosive process on the salt marsh and favors the formation of tidal creeks. To study it, a Cellular Automata model is proposed, based on the laws deduced from the observation of these phenomena in the field, and then verified with measurable data within macroscale time units. Therefore, the objective of this article is to model how the interaction between C. granulatus and S. perennis modifies the landscape of the salt marsh and influences the path of tidal creeks. The model copies the basic laws that rule the problem based on purely biological factors.The Cellular Automata model proved capable of reproducing the effects of the interaction between plants and crabs in the salt marsh. A study of the water drainage of the basins showed that this interaction does indeed modify the development of tidal creeks. Model dynamics would likewise follow different laws, which would provide a different formula for the probability of patch dilation. The patch shape can be obtained changing the pattern that dilates.     

Investigating the land use characteristics of urban integration based on remote sensing data: experience from Guangzhou and Foshan

Abstract

Urban integration is an increasingly popular phenomenon. It is important to characterize the dynamic processes of urban integration from the perspective of geography. However, previous studies mainly focussed on the overall characteristics of a single entity and failed to consider inter-city influences. Therefore, this study aims to systematically investigate this phenomenon. Firstly, spatial clustering technique and landscape metric were used to analyze the temporal change of spatial layout. Secondly, we built multiple ring buffers of city boundary to reveal the directional differences in urban expansion. Lastly, we compared the observed land use data with the results simulated by cellular automata model. These experiments have shown that the urban expansion characteristics during the urban integration process are rather different from those of a single city, and the above methodology can effectively characterize the urban integration phenomenon. The successful example of Guangzhou–Foshan integration could also provide practical experience for other similar attempts.     

人工免疫系统与嵌入规划目标的城市模拟及应用

人工免疫系统(AIS) 具有强有力的计算能力, 可以通过免疫识别、克隆选择、免疫学 习、免疫记忆等功能来进行模式识别和自适应学习。AIS 所具有的自学习、自适应和记忆的能力非常适合于复杂地理过程的研究。而元胞自动机(CA) 是研究复杂系统非常方便和有效的工具。将人工免疫系统和元胞自动机相结合, 建立了城市演变的模拟和规划模型。该模型通过改变抗体的进化变异机制, 把规划目标嵌入到AIS 算法中, 抗体将会逐渐朝着规划目标“进化”, 从而模拟出基于不同规划情景的城市发展空间格局, 为城市和土地利用规划提供决 策支持。设计了6 种不同的城市发展方案, 利用该模型模拟了不同规划方案下珠江三角洲城市的发展情景(1988-2002 年)。并比较了不同模拟情景结果城市的紧凑性: “城市中心” 和 “城市中心-高速公路”发展模式的城市形态更为紧凑, 破碎度较低; 而“镇中心” 和“道路”发展模式形成的城市形态则比较凌乱和分散。模拟结果和分析表明: “城市中心-高速 公路”是珠江三角洲最适合的城市发展模式。