Observer elevation changes the predicted view
The starting surface determines the observer's elevation in a viewshed calculation.
A digital terrain model (DTM) follows the ground, while a digital surface model (DSM) includes trees and buildings. Using the DSM to support the observer can raise the eye elevation and change which obstructions the sightline clears.
See the two layersWe tested both starting surfaces with the DSM checking what blocked the view. About 1 in 80 places we kept in the test gave a different answer.
The overall results for the public properties we tested stayed the same.
A viewshed maps where the model predicts you can see something. For a person on the ground, put their feet on the DTM and check obstructions with the DSM.
One place, two starting surfaces
A line of sight runs from your eyes to the target, the thing you're trying to see. The higher start clears the obstacle in this example.
The supporting surface is where the model puts your feet. Observer height is how far your eyes sit above it: 1.8 m in both versions.
Eyes placed higher
Surface-supported · DSM · Clear view · +0.94 m
Starting on the DSM raises the observer by 1.15 m. That lift takes the line of sight above the obstacle.
Feet on the ground
Ground-supported · DTM · Blocked view · −0.10 m
Starting on the DTM keeps the observer's feet on the ground. The lower line of sight hits the obstacle.
Clearance is the line's height above or below an obstacle. Its lowest value is the minimum clearance: above zero clears; below zero is blocked.
Check the predicted view in person.
Read the studyTry a tree-and-turbine example
See the two elevation layers
This constructed example places the eyes 1.8 m above the chosen starting surface. The DSM at the observer is 0.4 m above the DTM. The tree is 100 m away; the selected upper blade tip is 180 m high and 1 km away.
Layers at the tree
The DTM follows the ground beneath the tree. The DSM rises over it. These panels use the same distance and height scales so you can compare the two layers at the same point.
Each panel shows 90 to 110 m from the observer and heights from 0 to 22 m. Horizontal and vertical scales differ.
| Model assumptions | Result |
|---|---|
| Terrain only: Feet on DTM; eyes 1.8 m above ground. Trees omitted. Tree clearance not tested | Tip visible |
| Surface-supported: Feet on DSM (+0.4 m); eyes 2.2 m above ground. DSM checks obstructions. Line at tree 19.98 m · clearance +0.18 m | Tip visible |
| Ground-supported: Feet on DTM; eyes 1.8 m above ground. DSM checks obstructions. Line at tree 19.62 m · clearance −0.18 m | Tip blocked |
At 19.5 m, all three report the tip visible. Ground support clears the tree by 0.12 m; surface support clears it by 0.48 m. Terrain only still leaves the tree out.
This is a constructed example. The close-up enlarges the tree detail; the test covers the selected upper blade tip.
The details and numbers
The two starting surfaces gave different results at 1,674 of 133,286 locations that passed the height filter, or about 1 in 80. Both versions checked the same DSM for obstructions. The overall classifications of the tested public properties stayed the same.
We excluded places where the DSM was at least 1.8288 m (six international feet) above the DTM. At the places we kept, starting on the DSM still shifted eye elevation by the full difference between the surface and ground heights.
In the recorded example, observer height is 1.8 m above either starting surface. At the observer, the DSM is 1.1543960571289062 m above the DTM. Both endpoint tests clear. The minimum clearances occur at sample index 38 in the complete 374-position profile: +0.9415054321289062 m for the DSM start and −0.09528350830078125 m for the DTM start. The C kernel stops after 39 samples at its first blocker. No terrain-only result was supplied for this recorded example.
Check visibility and public access in the field before using a modeled location as a viewpoint. These results concern the modeled lines of sight. The comparison does not test every elevation change or establish changes to assessment boundaries or receptor inventories.
The turbine is a separate constructed example with flat ground, eyes 1.8 m above the starting surface, a 0.4 m surface offset and a fixed upper blade tip 180 m high and 1,000 m away. The hub height and rotor diameter are both 120 m. Only the foreground tree height changes. The model joins surface heights with straight segments and treats that surface as opaque, ignoring openings beneath crowns and light passing through leaves. The height marker measures the tree's top; it doesn't trace its crown. Curvature, refraction and Fresnel corrections are off.