In this presentation, Mr Joe Ortiz describes how he implemented and tested a Python script to calculate a DTM based on papers:
Adaptive algorithm for large scale dtm interpolation from lidar data for forestry applications in steep forested terrain, Almasi S. Maguya, Virpi Junttila, Tuomo Kauranne
Algorithm for Extracting Digital Terrain Models under Forest Canopy from Airborne LiDAR Data, Almasi S. Maguya, Virpi Junttila, Tuomo Kauranne
Digital Terrain Model (DTM) Extraction from LiDAR Point Clouds in Densely Forested Areas
1. Digital Terrain Model (DTM)
Extraction from Lidar Point
Clouds in densely forested
areas
JOE ORTIZ
2. Importance of the DTM
Basis for all biomass estimates: need DTM and canopy model
Must eliminate spikes.
Our data is not typical.
Full waveform data – return approximately at peak.
3. Chose to adapt existing algorithms
Adaptive algorithm for large scale
dtm interpolation from lidar data for
forestry applications in steep forested
terrain, Almasi S. Maguya, Virpi
Junttila, Tuomo Kauranne
Algorithm for Extracting Digital
Terrain Models under Forest Canopy
from Airborne LiDAR Data, Almasi S.
Maguya, Virpi Junttila, Tuomo
Kauranne
4. Transforming Lidar point cloud
Translation to move
origin.
Rotation by multiplying
each coordinate by 3D
rotation matrix.
7. Dividing tiles into cells
5m
5m
• Divide tile into cells of minimum size
5m.
• Assign each cell elevation as the
minimum elevation of the cloud points
inside the cell.
• Gives a grid of 20x20 data points for
each 100mx100m tile.
• Apply standard deviation filter to grid
of points.
• Gives grid of data points with gaps.
9. DTM of area 300x200m area
Tile size is 100x100m
All tiles chose
quadratic fit.
Resolution is 10m in
both x and y
direction. Defines
cell size in DTM.
10. DTM of area 300x200m area
Tile size is 60x70m
1 tile chose linear fit.
All the others chose
quadratic fit.
Resolution is 10m in
both x and y
direction.
11. Removing spikes
Spikes appear at edges of tiles usually due to poor extrapolation by the
cubic spline interpolation model.
Create DTMs using different tile sizes: ~ 60-120m
Take median of elevations of each cell (cell size given by desired
resolution) for each DTM
Hopefully produces spikeless final DTM!
12. Conclusion/ What next?
Algorithm works well on tested areas. Surface is relatively smooth and
hopefully will soon be spikeless.
Test over more areas of more complex terrain.
Compare to other DTM extraction algorithms.
Select best DTM extraction algorithm or combine elements of several
algorithms.
Make a biomass estimate!