Multi-representation of spatial data is one of the hot topics in modern GIS. This paper proposed a bi-hierarchical multi-scale model to represent large vector data with two hierarchies

namely spatial elements and vertex coordinates

so as to improve the visualization efficiency. Firstly spatial elements were taken as the minimal units and organized with a multi-scale index to describe the changes of quantity or property of elements brought with scale changes.Then vertex coordinates were regarded as the minimal units and tagged with scale hierarchy labels to describe the geometric changes arising from different scales. Under the support of PostgreSQL

this paper implemented the bi-hierarchical multi-scale model by extending the index and functions

and developed the searching algorithms. Finally

a 1:10

000 scale land use dataset was taken as an example to validate the model and searching algorithms.The results showed that the visualization and transmission efficiency of large vector data could achieve great improvement with the proposed model and searching algorithms.