Trees are an important feature that impacts the visual aspect of the area, as well as how shadows should be interpreted. The following article describes how we place trees in our scenes.
Raster source
The tree dataset we rely on comes in the form of a raster that describes the tree canopy height at a certain area. A visual representation of the raster, together with a satellite image of the same area is shown below
The brighter the color, the taller the tree canopy
We can clearly recognise the same tree areas from the satellite image
The same area from a View park shadows
From this source we are given answers to two questions
Which areas are filled with trees?
How tall are these trees?
It is important to understand that we can not parse out individual trees.
Tree placement methodology
As stated, from the source we do not know the position of individual trees, but in the 3d scene we still decide where individual trees are placed.
This is done in the following way.
We parse the pixels from the raster and for every pixel that contains a canopy height above 5m we place a tree of the corresponding height. It looks like this
The pixels from the raster are used as reference for placing trees of certain height
We quickly see an issue, the placement of the trees clearly follows the grid of the rasters pixels.
To avoid the grid like nature, we add a random offset between 0-5m in any direction, it looks like this
The random offset hides the grid like nature of the underlying dataset
Conclusion
Trees are added in a way to visualise general areas where trees are located with an accurate height rather than the absolute position of individual trees.