• 论文 •

### 地理信息系统可视化专题制图要素分级探讨

1. 清华大学建筑学院GIS实验室,北京100084
• 收稿日期:1997-05-09 修回日期:1998-08-23 出版日期:1998-12-15 发布日期:1998-12-15
• 作者简介:党安荣,男,一九六四年生于陕西佳县。一九八五年毕业于陕西师范大学地理系,获“地理学”学士学位;一九八八年获西北大学“地图学与遥感”专业硕士学位,同年分配到“煤炭部航测遥感局计算中心”从事GIS与CAD工作;一九九七年获中国科学院遥感应用研究所“地图学与地理信息系统”专业博士学位,目前在清华大学建筑学院“城市规划与设计”博士后流动站工作,主要从事基于RS和GIS的城市总体规划研究。已在《遥感学报》、《地理科学》、《中国图象图形学报》、《测绘通报》、《环境遥感》、《中国煤田地质》等专业刊物及ICA. FIG, ISPRS和CPGIS等国际会议上发表论文30余篇。专长于RS与GIS的应用研究、GIS成果的图形表达等

### STUDY ON THEMATIC MAPPING ELEMENT CLASSIFICATION OF GIS VISUALIZATION

Dang Among

1. GIS Lab. of Architecture College, Tsinghua University, Beijing 100084
• Received:1997-05-09 Revised:1998-08-23 Online:1998-12-15 Published:1998-12-15

Abstract: A series of classification quantitative standard have been worked out since the research of G. F. Jecks and M. R. C. Coulson (1963). However, the characteristics of statistical data has been considered so much that the mapping effect of classification scheme was hardly cared. In 1980s, mapping information content based on the entropy function was used to evaluate classifi-cation scheme which took the statistical precision and mapping effect into account at the same time. This article uses mapping classification information content to determine the visualization effect of GIS. A new formula of calculating classification information content is worked out based on the general mapping information concept.Using one of the general formulas of numerical sequence and progression, lots of classifi-cation schemes can be acquired with many sets of parameters (x,d,rBk) values within their do-main. This is the basic condition for classification optimization. However, the class intervals or class breaks of classification scheme changing in a single regularity from the minimum value to the maximum value of mapping element. And the change regularity of mapping element data it-self are not always so simple. The real situations are different from one to another and as com-plex as several types of changing rules composed together. Author proposes a new idea of "par-tial numerical sequence and progression classification method". The idea of this new method is that dividing mapping elements data series into several subseries at first; and then, using differ-ent classification method to acquire classification scheme for each subseries; finally, generalizing all of the subseries classification schemes to determine the final optimal classification scheme.

• P208