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。相比单结构的卷积神经网络CA模型,多结构卷积神经网络CA模型的稳定性和模拟结果准确性有所提升。     

Lesbian and Queer Sustainable Farmer Networks in the Midwest

Heteropatriarchy underpins contemporary U.S. agriculture, even within the alternative sector. This paper builds on the legacies of women farmers and farmers of color creating peer networks to circumvent heteropatriarchal hurdles by investigating how lesbian, bisexual, trans, and queer (LBTQ) sustainable farmers access human resources. If and how did the farmers encounter or resist heteropatriarchy in this process? Drawing on four years of ethnographic research with 40 LBTQ Midwest sustainable farmers, I argue that resources through government agencies, neighborhood farmers, and like-minded practitioners did not necessarily align with LBTQ farmers’ sustainable practices or queer identities. LBTQ farmers convened with others at the intersections of their queerness and sustainable practices formally, informally, and through the labor market to access human resources removed from heteropatriarchal domination. I conclude that LBTQ farmer networks bolster human resources in sustainable agriculture and conservation practices.     

经济地理学视角下创新网络与区域增长研究述评

创新是引领经济发展的第一动力,创新与区域增长之间的关系成为经济地理学者关注的重点领域。文章通过梳理近年来有关创新网络测度、创新网络与区域增长关系、创新网络作用于区域增长方式等方面的文献发现：网络资本可以弥补社会资本在解释企业创新结网经济价值方面的不足,区域增长呈现出网络化特征已成为学界共识,知识流与邻近性能较好地解释创新网络与区域增长的关系机理。然而,当前研究对社会资本如何促进区域增长,网络资本与区域增长关系模型如何构建,不同类型的邻近性与知识对区域增长的影响有何差异等问题尚不明确,建议重视网络资本对区域增长的作用并实证检验二者的关联,同时还应重视创新网络中企业家精神、创新个体心理行为特征等因素对区域增长的影响。     

结合深度学习和图割法的遥感影像建筑物检测

提出一种基于卷积神经网络和图割法的自动提取高分影像建筑物的方法。首先,通过卷积神经网络定位与检测建筑物的位置,逐一提取单个建筑物轮廓,利用检测结果分别建立建筑物和非建筑物的高斯混合模型（GMM）,然后结合最大流最小割的图像分割方式实现全局优化,完成建筑物初步提取,最后用形态学进行优化。通过试验证明了该方法的可行性。     

A surface network based method for studying urban hierarchies by night time light remote sensing data

Urban hierarchies are closely related to economic growth, urban planning and sustainable urban development. Due to the limited availability of reliable statistical data at fine scales, most existing studies on urban hierarchy characterization failed to capture the detailed urban spatial structure information. Previous studies have demonstrated that night time light data are correlated with many urban socio-economic indicators and hence can be used to characterize urban hierarchies. This paper presents a novel method for studying urban hierarchies from night time light data. Night time light data were first conceptualized as continuous mathematical surfaces, termed night time light surfaces. From the morphology of these surfaces the corresponding surface networks were derived. Hereafter, a night time light intensity (NTLI) graph was defined to describe the morphology of the surface network. Then, structural similarity between the night time light surfaces of any two different cities was calculated via a threshold-based maximum common induced graph searching algorithm. Finally, urban hierarchies were defined on the basis of the structural similarities between different cities. Using the 2015 annual NPP-VIIRS night time light data, the urban hierarchies of 32 major cities in China were successfully examined. The results are highly consistent with the reference urban hierarchies.     

长白山景区旅游安全风险动态评价研究

以长白山景区旅游安全为研究对象,以鱼骨图、动态贝叶斯、GIS技术等为基本研究方法,从研究区自然环境、社会环境及责任人为3个方面出发,筛选景区致险因子,构建景区旅游安全风险危险性评价指标体系,利用动态贝叶斯方法综合构建景区旅游安全风险动态评价模型;并以实测数据及景区统计数据为依据,划分景区旅游安全风险评价的4个动态时段,综合实现景区旅游安全风险动态风险评价。研究结果表明：① 中等以上风险区域呈条带状分布;② 高风险区域与主要景点重合;③ 长白山景区安全风险发生高概率的时段发生在第三个时段（12:00~14:00）;较高概率发生分别在第二个时段（10:00~12:00）与第四个时段（14:00~16:00）;中等概率发生较高的时段在第四个时段（14:00~16:00）;较低概率发生在第一个时段（8:00~10:00）。     

Geospatial data mining for digital raster mapping

We performed an in-depth literature survey to identify the most popular data mining approaches that have been applied for raster mapping of ecological parameters through the use of Geographic Information Systems (GIS) and remotely sensed data. Popular data mining approaches included decision trees or “data mining” trees which consist of regression and classification trees, random forests, neural networks, and support vector machines. The advantages of each data mining approach as well as approaches to avoid overfitting are subsequently discussed. We also provide suggestions and examples for the mapping of problematic variables or classes, future or historical projections, and avoidance of model bias. Finally, we address the separate issues of parallel processing, error mapping, and incorporation of “no data” values into modeling processes. Given the improved availability of digital spatial products and remote sensing products, data mining approaches combined with parallel processing potentials should greatly improve the quality and extent of ecological datasets.     

凸包插值算法的改进与实现

研究了凸包插值算法在n维欧拉空间构造Delaunay三角网的应用,提出了其三角剖分不惟一的重大缺陷,采用了限制区域生长法解决了这一缺陷。