Fig 1.
Location of the study area.
Fig 2.
Working flow of DEM terrain generalization.
Fig 3.
Schematic diagram of the tree structure creation.
Fig 4.
Example of a watershed tree for pruning.
Fig 5.
Schematic diagram of generating terrain profile lines in one direction.
Fig 6.
Schematic diagram of terrain profile line simplification.
Fig 7.
Boundary points traversal of the W8D algorithm.
Fig 8.
Boundary points traversal of the W8D algorithm.
Fig 9.
Spatial distribution of the terrain feature points under different thresholds.
The basic DEM image is original didn’t generalization and the dots in orange are the terrain feature points.
Fig 10.
Results of the generalized DEMs with different thresholds.
Table 1.
The accuracy assessment of the generated DEM against the 5 m resolution original DEM.
Fig 11.
A comparison of the mean difference for the 5 m resolution DEM under different thresholds.
The y-axis according to mean difference and error bars according to SE Mean in Table 1.
Table 2.
The accuracy assessment of the generated DEM against the 25 m resolution original DEM.
Fig 12.
A comparison of the mean difference for the 25 m resolution DEM under different thresholds.
The y-axis according to mean difference and error bars according to SE Mean in Table 2.
Fig 13.
The comparison diagram of the different algorithm-generated DEMs enhanced the local details.
Fig 14.
Overlapped contours extracted from the reconstructed DEMs.
The thin lines are the contours that are converted from the actual 1:50000 DEM, whereas the thick lines are converted from the 1:50000 DEM and generalized from the 1:10000 DEM through different methods. And the black arrows pointing to the regions indicate the examples of effectiveness of overlaps using difference algorithms.
Fig 15.
The comparison diagram of the distribution frequency of the elevation under 25 m resolution DEM.
Fig 16.
The comparison diagram of the slope frequency under 25 m resolution DEM.
Fig 17.
The radar graph of the aspect frequency.