A comprehensive framework for exploratory spatial data analysis: Moran location and variance scatterplots

Abstract

A significant Geographic Information Science (GIS) issue is closely related to spatial autocorrelation, a burning question in the phase of information extraction from the statistical analysis of georeferenced data. At present, spatial autocorrelation presents two types of measures: continuous and discrete. Is it possible to use Moran's I and the Moran scatterplot with continuous data? Is it possible to use the same methodology with discrete data? A particular and cumbersome problem is the choice of the spatial-neighborhood matrix (W) for points data. This paper addresses these issues by introducing the concept of covariogram contiguity, where each weight is based on the variogram model for that particular dataset: (1) the variogram, whose range equals the distance with the highest Moran I value, defines the weights for points separated by less than the estimated range and (2) weights equal zero for points widely separated from the variogram range considered. After the W matrix is computed, the Moran location scatterplot is created in an iterative process. In accordance with various lag distances, Moran's I is presented as a good search factor for the optimal neighborhood area. Uncertainty/transition regions are also emphasized. At the same time, a new Exploratory Spatial Data Analysis (ESDA) tool is developed, the Moran variance scatterplot, since the conventional Moran scatterplot is not sensitive to neighbor variance. This computer-mapping framework allows the study of spatial patterns, outliers, changeover areas, and trends in an ESDA process. All these tools were implemented in a free web e-Learning program for quantitative geographers called SAKWeb© (or, in the near future, myGeooffice.org).     

Interactive visual cluster detection in large geospatial datasets based on dynamic density volume visualization

The emerging spatial big data (e.g. detailed spatial trajectories, geo-referenced social media data) provide tremendous opportunities for GIScientists and geographers. However, their large volume also poses challenges to existing spatial data analytical techniques (including visual analytical techniques). This article presents an interactive visual approach to detect clusters from those emerging data sets based on dynamic density volume visualization in a three-dimensional space (two spatial dimensions plus a third temporal or thematic dimension of interest). Cluster can be visually discovered through dynamic adjustment of density to colour/opacity mapping and extracted through flexible selection tools. The approach was tested on a large simulated data-set and a spatial trajectory data-set. The results show that the approach can overcome the visual clotting problem in traditional visualization tools caused by large data volume and facilitate the involvement of domain knowledge in analysis. It can effectively support visual cluster detection in the emerging large geospatial data sets.     

遥感影像样本数据集研究综述

随着机器学习、深度学习等人工智能技术在遥感领域的不断应用与发展,基于海量样本的数据驱动模型已经成为遥感影像信息提取的一种新的研究范式,其对样本数据的规模、质量、多样性等提出了更高要求。最近,国内外众多学者和研究机构相继发布了一系列遥感影像样本数据集,为大数据时代下遥感影像的信息提取和智能解译等奠定了研究基础。然而目前尚缺乏对上述影像样本数据集的综合分析,针对这一问题,本文在文献检索与分析的基础上,归纳总结了124个具有一定影响力且应用广泛的遥感影像样本数据集并对其元数据进行了分析,并提供了数据来源、应用领域与关键词的发展变化,分析了数据集在空间、时间、光谱分辨率上的差异,以应用领域为依据将其划分为场景识别、土地覆被/利用分类、专题要素提取、变化检测、目标检测、语义分割等8个类别并以部分数据为例进行了具体分析,总结了深度学习模型在数据集上的研究进展,并针对稀疏样本导致的模型过拟合问题,探讨了样本时空迁移、小样本和零样本学习、样本主动发现、样本生成等在遥感影像信息提取中的应用前景。本文首次对遥感影像样本数据集进行了综述研究,可为相关领域科研人员提供数据参考。     

Developing local measures of spatial association for categorical data

This paper describes a procedure for extending local statistics to categorical spatial data. The approach is based on the notion that there are two fundamental characteristics of categorical spatial data; composition and configuration. Further, it is argued that, when considered locally, the latter should be measured conditionally with respect to the former. These ideas are developed for binary, gridded data. Local composition is measured by counting the numbers of cells of a particular type, while local configuration is measured by join counts. The approach is illustrated using a small, empirical data set and an ad hoc procedure is developed to deal with the impact of global spatial autocorrelation on the local statistics.The author gratefully acknowledges financial support from the GEOIDE Network of Centres of Excellence (ENV #4) and the helpful comments of three anonymous reviewers.     

Hierarchical Bayesian spatial models for predicting multiple forest variables using waveform LiDAR,hyperspectral imagery,and large inventory datasets

In this paper we detail a multivariate spatial regression model that couples LiDAR, hyperspectral and forest inventory data to predict forest outcome variables at a high spatial resolution. The proposed model is used to analyze forest inventory data collected on the US Forest Service Penobscot Experimental Forest (PEF), ME, USA. In addition to helping meet the regression model's assumptions, results from the PEF analysis suggest that the addition of multivariate spatial random effects improves model fit and predictive ability, compared with two commonly applied modeling approaches. This improvement results from explicitly modeling the covariation among forest outcome variables and spatial dependence among observations through the random effects. Direct application of such multivariate models to even moderately large datasets is often computationally infeasible because of cubic order matrix algorithms involved in estimation. We apply a spatial dimension reduction technique to help overcome this computational hurdle without sacrificing richness in modeling.     

Web-based analytical tools for the exploration of spatial data

This paper deals with the extension of internet-based geographic information systems with functionality for exploratory spatial data analysis (esda). The specific focus is on methods to identify and visualize outliers in maps for rates or proportions. Three sets of methods are included: extreme value maps, smoothed rate maps and the Moran scatterplot. The implementation is carried out by means of a collection of Java classes to extend the Geotools open source mapping software toolkit. The web based spatial analysis tools are illustrated with applications to the study of homicide rates and cancer rates in U.S. counties.This research was supported in part by a number of grants from the US National Science Foundation: NSF Grant SBR-9410612, BCS-9978058, to the Center for Spatially Integrated Social Science (csiss), and a grant from the National Consortium on Violence Research (ncovr is supported under grant SBR-9513040 from the National Science Foundation). In addition, support was provided by grant RO1 CA 95949-01 from the National Cancer Institute. Special thanks to Dr. Eugene J. Lengerich of the Pennsylvania State Cancer Institute for providing the data on colon cancer diagnoses.     

地球化学的空间自相关异常信息提取方法

针对地球化学数据存在的空间分布相关性特征,该文提出了一种基于空间自相关统计的地球化学异常提取方法。以内蒙古浩布高矿床外围的土壤地球化学数据为例,通过对Sn、Cu元素地球化学数据在不同空间间隔上的全局自相关计算,测算其空间聚集的程度,选取聚集程度最高时的间隔距离作为局部空间自相关的参数,通过局部Moran’s I值研究元素的空间分布特征,分析空间聚类和异常值,从而提取地球化学异常。结果表明,局部空间自相关分析可以揭示Sn、Cu地球化学数据的空间分布特征,能够更好地提取地球化学弱缓异常,说明空间自相关是一种有效的地球化学异常识别方法。     

主元分析变换空间上的鉴别共同矢量人脸识别方法

给出了在主元分析(PCA)变换空间上求取DCV投影矩阵的方法(PCA+DCV),在保留所有鉴别信息的条件下,显著降低了算法复杂度,提高了运算效率。进一步提出了依据主元成分对应特征值进行适度权重的DCV识别方法(WPCA+DCV),一定程度上减小因光照、饰物遮挡等造成的面部变化带来的识别影响,增强表征信息,提高识别率。在ORL、YALE和AR人脸库上的实验结果证实了本方法的性能。     

基于核密度估计的广佛都市区路网演变分析

道路密度信息广泛应用于景观分析、路网规划、城市边界提取与道路网综合等领域。提出了一种道路密度分区方法,该方法首先生成道路交叉点和端点的Voronoi图,然后运用G_i^*提取局部的Voronoi单元面积的高值聚集区与低值聚集区,最后将95%置信度下的高值区域和低值区域对应的相邻Voronoi单元合并。对Google地图上14级、13级和12级香港的道路网进行了实证研究,结果表明,该方法生成的密度分区大致反映了道路网的疏密,并较好地反映了选取前后各区域道路密度的对比规律。将本文方法与网格密度法进行对比,结果表明本文方法优于网格密度法。     

Change detection thresholds for remotely sensed images

 The detection of change in remotely sensed images is often carried out by designating a threshold to distinguish between areas of change and areas of no change. The choice of threshold is often arbitrary however. The purpose of this paper is to offer a statistical framework for the selection of thresholds. The framework accounts for the facts that one carries out multiple tests of the null hypothesis of no change, when searching for regions of change over an image with a large number of pixels. Special attention is given to global spatial autocorrelation, which can affect the selection of appropriate threshold values. Received: 8 March 2001 / Accepted: 12 October 2001     

道路网与兴趣点相结合的城市中心提取方法

为空间数据添加接近人们思维以及适宜认知的高阶信息是改善其可用性的重要途径。城市中心是这一类信息的典型案例,它在人们的社会活动中具有重要作用。本文提出一种单纯运用道路网和兴趣点提取城市中心的方法。该方法首先运用G*i提取了路网的密集区域,确定了包含城市中心的大致区域;然后根据该区域中特定类型兴趣点的网络核密度确定了城市中心的精确范围。对英国利物浦、加拿大多伦多和巴西库里蒂巴进行了试验,查准率为0.74~0.8,查全率为0.53~0.67,结果表明该方法能较为有效地提取城市中心。对方法的两个关键影响因素:G*i的距离测度以及网络核密度的带宽进行了敏感性分析,固定距离法为合适的距离测度方法,而600~900m为适宜带宽。     

An oracle-based data management method for large database in CyberCity GIS

An Oracle8i-based approach is proposed to manage the integrated databases of large CyberCity. This approach consists of three schemes: ? a special R+-tree index is designed to accelerate spatial retrieving, in which the bounding boxes of local regions have no intersection and all leaf nodes of the R+-tree (geometry records) have no repetition; ∪ different data compression algorithms are adopted to compress the digital elevation models, 3D vector models and images, such as LZ77 lossless compression algorithm for compression of vector data and JPEG compression algorithms for texture images; ? in order to communicate with Oracle8i database, a CyberCity GIS spatial database engine (SDE) is designed. On the basis of this SDE prototype a case study is done.     

An oracle-based data management method for large database in CyberCity GIS

An Oracle8i-based approach is proposed to manage the integrated databases of large CyberCity. This approach consists of threeschemes: ① a special R -tree index is designed to accelerate spatial retrieving, in which the bounding boxes of local regions have no intersection and all leaf nodes of the R -tree (geometry records ) have no repetitiont;② different data compression algorithms are adopted to compress the digital elevation models, 3D vector models and images, such as LZ77 lossless compression algorithm for compression of vector data and JPEG compression algorithms for texture images;③ in order to communicate with Oracle8i database, a CyberCity GIS spatial database engine (SDE) is designed. On the basis of this SDE prototype a case study is done.     

基于Hilbert排列码与R树的海量LIDAR点云索引

本文分析当前索引方法存在问题,针对高效海量点云数据的要求,提出一种基于Hilbert码与R树的二级索引方法。论文阐述了二级索引的建立原理与方法,可通过聚类方法与R树度M值来的优化第一级索引;使用Hilbert R树作为第二索引,可以有效控制两级R树的高度,同时点云的增加与更新可只在局部进行。最后本文通过两组实验来验证该数据组织方法的可行性和跟其他索引(KD树与四叉树)进行比较,得出它是一种高效管理海量点云的方法。     

一种非凸包边界约束不规则三角网生成算法

为研究非凸空间离散数据的空间划分,建立了耦合数据集的凸凹性、数据规模、离散度的空间划分的数学模型。利用Laves划分标识[3⁶]对非凸空间离散数据进行有限区域变比例划分,然后通过地形曲面微分单元与数据规模的偏导函数关系,耦合离散度计算空间划分的单元间距和数量。最后通过构建DEM,可视化验证和对比分析发现,耦合模型能够计算出非凸离散空间数据空间划分单元的间距和数量,也能实现不同分辨率的划分单元的无缝拼接;且当试验数据从110组递增至440组时,该模型耗时仅是[4⁴]标识划分和Delaunay的1/10～1/3,且随数据规模成倍增加时耗时基本呈线性增长,收敛性较好,但耗时随离散度增加而增长。     

顾及地价空间分布规律的城市基准地价以价定级方法研究

提出了一种城市基准地价以价定级方法。该方法用矢栅混合模型划分评价单元,对地价样点进行趋势面分析筛选,使用最短路径距离代替欧氏距离对地价样点进行空间插值,叠加分析后得到各评估单元的指标地价,用其代替传统的因素综合得分对基准地价评估区域分等定级,同时测算出各级别基准地价。以上海市为例对该方法的实用性进行了验证。