Abstract: Objective The complexity of geometry-based methods for integrating vector data and DEM grids seri-ously limits their application on large area.To overcome this deficiency,a novel“drift”algorithm isput forward and designed based on the idea of grid cell decomposition and the partition model of thespherical Degenerate Quadtree Grid(DQG).An experiment was developed in which USGS GTOPO30DEM data,ASTER GDEM data and the 1∶1 000 000 DCW roads data sets for the China andChongqing area were used test the proposed drift algorithm.Tests show that the the“drift”algorithmis only 25% of the traditional one,and the mean error of the drift operation on vector nodes can becontrolled within about half a DQG cell.Moreover,the phenomenon of vector polyline going acrossthe multi-scale DEM grids would are avoided completely.