It has been widely recognized that, in order to automate the map generalization process, cartographic knowledge needs to be formalized. One method of knowledge acquisition is to analyse existing maps. Previous studies have concentrated on describing the phenomena on the surface. A typical example is the percentage change of the number of symbols or the percentage of open space on smaller scale maps. This study aims to go one step further, i.e. to analyse the association such changes with thematic attributes. In this study, road features on topographic maps of Hong Kong from 1: 1000 to 1:200 000 are studied; six types of thematic attributes, i.e. 'type', 'length', 'width', 'number of lanes', 'number of traffic ways' and 'connectivity', are considered; and two statistic parameters, Lambda and Somers' Delta, are employed. It has been discovered that the dependence of road elimination on these attributes occurs in the following order: type (0.73), length (0.53), number of lanes (0.48), number of traffic directions (0.41), width (0.36) and connectivity (0.19). If these numbers are normalized into percentages, then these values become: type (27%), length (20%), number of lanes (18%), number of traffic directions (15%), width (13%) and connectivity (7%). Such results could be used to formulate an overall weight to determine whether a particular road should be deleted, merged or combined in the generalization process, in order to retain a certain proportion of roads at a smaller scale.
ASJC Scopus subject areas
- Earth-Surface Processes