River network dynamic selection with spatial data and attribute data based on rough set

This paper considers the river network generalization problem by taking advantage of the rough set theory. Through a detailed analysis and weighing of the various spatial as well as attribute data that determine the importance of rivers, some of the data are selected as the conditional attributes to construct the decision table of a rough set. Then the rough set theory is employed to evaluate the importance of rivers. By doing so, a dynamic selection method is putted into effect to achieve the representation of river network in different scales under the condition of maintaining its spatial features. The experimental results demonstrate the validity of the method and in particular, the consistency of the spatial features of the river networks after the generalization procedure.