面向空气质量监测数据时空多维属性的可视分析方法

发布时间：2018-03-20 22:42

本文选题：大气污染　切入点：可视分析　出处：《计算机辅助设计与图形学学报》2017年08期 　论文类型：期刊论文

【摘要】：空气质量监测数据具有显著的时空多维属性.传统的平行坐标技术虽然能够有效地展示数据的多维属性信息,却在分析与解读空气质量监测数据各个属性的时空变化规律方面表现出一定的局限性,在平行坐标展示空气质量监测数据多维属性的基础上,提出一种支持用户交互式探索大气污染时空特征的可视分析方法.首先利用平行坐标展示空气质量监测数据,支持用户交互改变时空维度以及指定坐标轴排列顺序;然后引入角度面积正负相关性等方式度量数据在平行坐标系中的布局差异,并且通过矩阵图和交互式柱状图分别展示不同时空维度下数据的布局差异;再综合考虑各个属性之间的数据布局差异,构建相似性矩阵,利用多维标度法对当前时空维度的数据进行降维,获得初始数据在低维空间的表示;最后利用层次聚类方法对低维空间的数据表达做聚类分析,并且分别设计时钟隐喻图和地域抽象图描述各个类别的时空节点组成.集成上述可视化算法设计便捷的用户交互模式,开发面向空气质量监测数据时空多维属性的可视分析原型系统,为用户快速分析和解读大气污染的时空特征及潜在规律提供有效手段.通过大量的可视化效果及用户反馈结果,进一步验证了文中所提可视分析方法的有效性和实用性.
[Abstract]:Although the traditional parallel coordinate technology can effectively display the multidimensional attribute information of the data, However, there are some limitations in analyzing and interpreting the temporal and spatial variation of the air quality monitoring data. On the basis of displaying the multidimensional attributes of the air quality monitoring data in parallel coordinates, In this paper, a visual analysis method is proposed to support the interactive exploration of spatio-temporal characteristics of air pollution. Firstly, the air quality monitoring data are displayed in parallel coordinates to support the users to interactively change the spatial and temporal dimensions and specify the coordinate axis order. Then we introduce the positive and negative correlation of angle area to measure the layout difference of data in parallel coordinate system, and show the difference of data layout in different space-time dimension by matrix graph and interactive histogram. Then considering the data layout difference among the attributes synthetically, the similarity matrix is constructed, and the dimensionality of the current space-time dimension data is reduced by using the multidimensional scaling method, and the initial data representation in the low-dimensional space is obtained. Finally, the hierarchical clustering method is used to analyze the data representation in low dimensional space. And the clock metaphor map and the region abstract graph are designed to describe the composition of each class of space-time nodes, and the user interaction mode is designed by integrating the above visualization algorithms. A visual analysis prototype system for spatiotemporal and multidimensional attributes of air quality monitoring data is developed, which provides an effective means for users to quickly analyze and interpret the space-time characteristics and potential laws of air pollution, and through a large number of visualization effects and user feedback results, The effectiveness and practicability of the visual analysis method proposed in this paper are further verified.
【作者单位】：浙江财经大学信息学院;浙江大学CAD&CG国家重点实验室;浙江财经大学应用经济学博士后流动站;浙江工商大学统计与数学学院;
【基金】：国家自然科学基金(61303133,61472354) 全国统计科学研究项目(2015LD03) 浙江省科技厅公益技术应用研究计划项目(2014C31057) 教育部博士后科学基金(2015M571846) 浙江省自然科学基金(LY15F020014) 浙江省一流学科A类(浙江财经大学统计学)2016年度规划项目浙江大学CAD&CG国家重点实验室开放课题(A1417)
【分类号】：X831

【相似文献】