Transcript LiDAR

What should be on your radar…
Current research
directions for
enhanced forest
inventory.
Doug Pitt
CFRU Workshop
2015
Orono, Jan. 22
Overview…
Changes in cost:benefit
Point clouds from photos
Terrestrial LiDAR
Species prediction
Substrate prediction
Growing the inventory
Fibre attribute prediction
Alternative platforms
Habitat & ecological indicators
Increased resolution; lower costs
Despite rapid changes in technology, you
will never be sorry for a WTW LiDAR
acquisition today…
M. Woods
M. Woods
LiDAR 2005
M. Woods
LiDAR 2012
- Remote Sensing 4: 830-848
Signal waveform
Point clouds from photos through
photogrammetric pixel
matching…
M. Woods
- Forests 4: 518-536
LiDAR DTM + Photos for updates?
Photo point cloud
Height (m)
LiDAR point cloud
Mean: 7.95 m
Std_Dev: 6.66 m
P50: 5.08 m
P90: 17.85 m
M. Woods
Mean: 16.15m
Std_Dev: 2.12 m
P50: 15.98 m
P90: 18.89 m
LiDAR DTM + Photos for updates?
Ave. DBH (cm)
BA(m²/ha)
PHOTO
30
Sawlog vol. (m³/ha)
50
250
40
200
30
150
20
100
10
50
25
20
15
10
10
15
20
25
30
0
0
10
20
30
40
50
0
0
50
100
150
LiDAR
- Can. J. Rem. Sens. 40(3): 214-232
200
250
Terrestrial LiDAR…
J-F Cote
L-Architect…
(LiDAR data to tree architecture)
FPOptitek
- Env. Modelling and Software 26: 761-777
- Agr. and Forest Meteorology 166: 72-85
- Remote Sensing 39(S1): 41-59
Species classification (ITC)
M. Woods
- P.E.R.S. 72: 1287-1297
- PCI software package
…from LiDAR ?
~ 1 return /m2…
20m
20m
- Woods et al. (in prep.)
Actual (%)
n = 346 plots; 86 used for validation:
Predicted (%)
H
HC
H: 91 4
HC:
100
CH: 9
9
C:
4
CH
4
C
82
96
… with higher resolutions
M. Woods
- Segma (Benoit St-Onge, UQAM)
Substrate prediction
WAM
http://www.youtube.com/watch?v=LsnB9vV-QEY
…building on WAM?
Add slope, aspect, species preference…
…to predict soil texture?
@ 10-m pixel
resolution –
overall
accuracy 79%
- Geoderma 13-24: 239-240
Environmental input variables:
• Elevation (10m)
• Slope (%)
• Surface shape (Curvature)
• Mode of deposition (NOEGTS)
• Landcover
• Slope Position from TPI
• (macro window = 1km)
• (medium window = 500m)
• (micro window = 20m)
• Wetness Index
Growing the inventory
Site Index
D. Pitt
Age
Height
D. Pitt
…fibre attribute prediction?
Predicted Sb wood density (kg/m3)
700
600
500
400
400
500
600
700
Observed Sb wood density (kg/m3)
D. Pitt
- Forest Science 59: 231-242
- Canadian J. Forest Research 44: 416-475
Alternative platforms…
NHRI
Institut de recherche sur les feuillus nordiques
Northern Hardwoods Research Institute
D. Cormier
[email protected]
[email protected]
Sustainability; Habitat / indicators
D. Pitt
D. Pitt
D. Pitt
[email protected]
The future is exciting…
D. Pitt
[email protected]
…aiding photo interp.?
LiDAR Predicted GMV Raster
Use LiDAR rasters to create polygons?
Better DTM!
M. Woods
Ontario Base Map (20-m)
If there’s one good
reason to invest in
LiDAR, this is it!
LiDAR (1-m)
So what is LiDAR?
Rapid pulses of laser light
Sensor records return times, converted to
distances
GPS= x, y, z
Multiple returns
1st
return
2nd
Last
J. White
D. Pitt
Basic products
Digital Surface Model (DSM)
Digital Terrain Model (DTM)
B. St. Onge
Canopy Height Model (CHM)
LiDAR for EFI
height
Point clouds to inventory predictions...
D(%)
M. Woods
LiDAR for EFI
Predicted stand volume
- total (m3/ha)
Predicted
Inventory attributes (e.g., GTV) are related to
these point-cloud statistics...
350
Sb GTV (m3/ha) = 31.46 + 1.78(mean · p90)
300
250
200
tvol
150
1:1
100
50
0
0
50
100
150
200
250
300
350
Actual
Actual stand volume - total (m3/ha)
M. Penner