Research on Entropy Uncertainty Band of Linear Segments in GIS
Da Li · 2002
Spatial data uncertainty can directly affect the quality of digital products and GIS based decision making and is one of the important topics of geographic information system(GIS).Among them, positional uncertainty of linear segments is a research focus. Much attention and effort have been devoted to the modeling of positional uncertainty of line segments and a lot of models are put forward by scholars, such as e band,e band, g band and so on. Although the most of exist models could been divided into two classes,namely error band model and confidence region model, they are in nature two different form confidence region, which have different band width with respect to confident levels. Diversity of models makes difficult to visualize uncertainty. In order to overcome the limitation of above models, it is necessary to put forward entirely determinate uncertainty model for line segments. In this paper, by introducing the concept of information entropy theory, entropy error ellipse of two dimension random point and entropy error ellipsoid of three dimension random point are presented and probability of falling into them is computed respectively, and then point entropy uncertainty model is extended to entropy uncertainty band of line segments. Finally, some conclusion is drawn as follows:①Entropy uncertainty band for line segment is a mean uncertainty region,and is solely determined by union entropy and it is independent of a certain confident levels.②For normal distribution, entropy uncertainty band width for two dimension line segment is as 2.332 times as g band, and for three dimension line segment, as 2 564 times as g band.③Entropy uncertainty band is a very suitable measure index of linear positional uncertainty, because it has a determinate band width.