Application of Airborne LiDAR Measurement to Forestryarchive.sokugikyo.or.jp/pdf/gift/2laser/02.pdfApplication of Airborne LiDAR Measurement to Forestry 산림분야항공기LiDAR

Application of Airborne LiDAR Measurement to Forestry산림분야항공기LiDAR 계측의응용

森林分野への航空機LiDAR計測の応用

Yasumasa HirataForestry and Forest Products Research Institute (FFPRI)

산림종합연구소히라타야스마사

森林総合研究所平田泰雅

Forestry and Forest Products Research Institute

What parameters are required to measure in forest sector?

• Forest manager– Individual tree information– Stand height (Site quality)– Stand volume

• Forest officer– Stand volume– Stand condition

• Government officer– Total volume– Total carbon stock


木木

Intensity

First return

Last return

One shot of laser

GPSGPS

IMUIMU

TimeTime

Tim

e

first return

last return

Forest measurement by ALS

• Digital surface model (DSM) is created from first return.

• Digital elevation model (DEM) is created from last return.

• Digital canopy height model (DCHM) =DSM –DEM

• Other information (waveform, penetration rate, etc.) is also used to estimate stand condition.


What can be measured using ALS?

• Direct measurement– Tree height

– Crown area

– Crown depth

• Estimated parameter– DBH

– Volume

– (Carbon stock)

height

Crown area

DBH

Crown depth


0

20

40

60

80

0 10 20 30 40 50

0

20

40

60

80

0 10 20 30 40 50

Chamaecyparis obtusa

• Local maximum filter is applied toDCHM to extract individual tree-tops.

• Watershed method or valley-followingmethod is applied to DCHM to extractindividual crowns.

Cryptomeria japonica

Crown area(m2)

DBH

（cm

）胸

高直

径（cm

）

Extraction of tree height, crown area and tree number from DCHM

R2=0.82

R2=0.71Extraction and estimation

Forestry and Forest Products Research InstituteHirata et al. (2009)

Measuement in a Japanese cedar pantation

V (m/sec)

H (m)δ

γ

ｎ (Hz/sec)

F (Hz/sec)

θ

δ

300 m600 m1200 m

24 deg.

25,000 Hz

14 m/sec

1.0 mrad

34 Hz

Measurement Parameters

Lightthinning

Heavythinning

Nothinning

99 trees

111 trees125 trees

118 trees

88 trees

113 trees

0.15ha 0.15ha0.20ha

0.15ha 0.15ha0.20ha

Heavythinning

Lightthinning

Nothinning

50m

50m

30m 30m40m

Plot distribution


3-D display of DSM and DEM

DSM

DEM

Results of ALS measurement

DCHM


3-D model

Hirata (2005)

00 40

40

H ALS(m)H

grou

nd (m

)

H ground = 1.009 * H ALS – 0.59r2 = 0.85

Tree height from DCHM

Extraction of individual trees

thinning operation

number of standing trees

number of extractive trees

extractive rate (%)

heavy (0.3 ha) 142 136 95.8

light (0.4 ha) 245 212 86.5

no (0.3 ha) 270 203 75.2

Total(1.0 ha) 657 551 83.9

Tree number and height from DCHM

Forestry and Forest Products Research InstituteHirata (2005)

Hf Hl Hl HfHf Hl Hl Hf

Factors of underestimation and overestimation of tree height in mountainous terrain


• Trees usually incline to one side in mountainous area.• 70 % of trees inclined to the lower side of slope.• Average of (H l – H f) was 0.2 meter.

Hirata (2004)

Relationship between actual tree height and measured tree height (slope=30º)

Valley sideHill sideOn contour

Inclination of standing tree (degree)


25 points/m22 – 3 points/m2

In the case of low sampling density, some tree-tops are not involved by any footprints.

In the case of high sampling density, each tree-top is involved by one of footprints.

crown

footprint

Sampling density


sampling density = 22.53 sampling density = 11.28

sampling density = 5.64 sampling density = 2.82 sampling density = 1.41

Sampling density

sampling density = 0.35sampling density = 0.71

DCM derived from different sampling density

sampling density = 22.53 sampling density = 11.28

sampling density = 5.64 sampling density = 2.82 sampling density = 1.41

sampling density = 0.35sampling density = 0.71

As sampling density runs lower, extractive tree-tops decrease.

Simulation of sampling density

Data size Mean sampling density (points/m2)

S. D. (points/m2)

Original 22.53 12.43

1/2 11.28 6.37

1/4 5.64 3.48

1/8 2.82 2.06

1/16 1.41 1.31

1/32 0.71 0.86

1/64 0.35 0.59

Generating dataset of different sampling density

0

10

20

30

40

50

60

70

80

90

100

0.0 5.0 10.0 15.0 20.0 25.0

25cm

50cm

1m

2m

Rate of involving more than 1 pulse data in every mesh size

sampling density (points/m2)

(%)


0

100

200

300

400

500

600

0 5 10 15 20 25

0.25m

0.5m

1mNum

ber

of

ext

ractive

tre

es

Sampling density of laser beams (points/m2)

mesh size

The relationship between sampling density and number of extractive trees

• In case of the sampling density of more than 5 points/m2, the rate of extractive trees with 1 meter and 0.5 meter mesh sizes against 0.25 meter mesh size were about 60 % and 90 % respectively.

• The rate of extraction of treetops from DCM declined suddenly in case the sampling density was below 3 - 5 points/m2.


30.0

30.5

31.0

31.5

32.0

0 5 10 15 20 25

0.25m

0.5m

1m

Sampling density of laser beams (points/m2)

Mean

tre

e h

eig

ht

of

ext

ractive

tre

es

(m)

mesh size

The relationship between sampling density and mean height of extractive trees in each mesh size

• The mean tree height in the same sampling density is smallest for the 0.25-meter mesh size.

• The difference between mean tree height by mesh size was 0.4 - 0.6 meter.

• The estimated height with low sampling density was underestimated.


truetrue

largerfootprint

largerfootprint

smallerfootprint

smallerfootprint

Difference of DEM and DSM by footprint size

Surface elevation

true point> smaller footprint> larger footprint

Ground elevation

larger footprint>= smaller footprint>= true point

In mountainous area


Coverage of canopy surface by footprint size

footprint = 0.3 m (sampling density = 24.8 /m2)

footprint = 0.6 m(sampling density = 10.1 /m2) footprint = 1.2 m(sampling density = 7.5 /m2)

DCM

Last return of ALS by footprint size

footprint = 0.3 m footprint = 1.2 mfootprint = 0.6 m


DCM derived from different footprint size

footprint = 0.3 m footprint = 1.2 mfootprint = 0.6 m

DCM0.3 DCM0.6 DCM1.2


0

100

200

300

400

500

-10 -8 -6 -4 -2 0 2 4 6 8 10

Difference (m)

Are

a (m

2)

Subtraction of DCHM by footprint size of 0.3 meter from DCHM by footprint size of 1.2 meters

DCHM1.2 – DCHM0.3


0

200

400

600

800

1000

-10 -8 -6 -4 -2 0 2 4 6 8 10

Difference (m)

Are

a (m

2)

Subtraction of DCHM by footprint size of 0.3 meter from DCHM by footprint size of 0.6 meter

DCHM0.6 – DCHM0.3


Penetration of laser through canopy


Penetration of laser in a deciduous forest

more than 50 species- Quercus serrata Murray- Fagus japonica Maxim.- Fagus crenata Blume


AcquisitionDate

Flight Speed Flight Altitude

Beam divergence

Pulse Frequency

Sampling Density

24/08/2001 13.9 m/sec 250 m 1.0 mrad 25,000 Hz 25.0 pts/m2

14/04/2002 13.9 m/sec 300 m 1.0 mrad 25,000 Hz 31.9 pts/m2

in the leafless seasonin the full‐leaf season

0

100

%

total penetration rate 20% total transmittance 69%

Penetration in different seasons



Last return of laser in different seasons



First return of laser in different seasons



First and last return in cross section in different seasons



Last return in cross section in different seasons


Evaluation of different levels of thinning


Relative frequency distributions of the heights above the ground where laser pulses reached


Thinning and penetration rate


Total basal area of remaining trees and penetration rate



Stnand parameter by ALS

Hirata et al. (2008)

Stand density from DCHM


Mean tree height from DCHM


Stand volume by ALS


Concluding remarks

• What parameters are required to measure in forest sector?– Depend on users and purposes

• Measurement by ALS effects on– Sampling density– Footprint size– Stand density– Species– Topography– Inclination

• Waveform laser scanning• Cost and archieve for the

expansion of use


References of this presentation• Hirata, Y., 2004. The effects of footprint size and sampling density of

airborne laser scanning to extract individual trees in mountainous terrain. International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences Vol. XXXVI-8/W2, 102-107.

• Hirata, Y., 2005. Influence of transmittance and sampling density of laser beams in forest measurement of a Cryptomeria japonica stand with an airborne laser scanner (in Japanese with English abstract). Japanese Journal of Forest Planning 39, 81-95.

• Hirata, Y., 2007 Forest Measurement with Airborne Laser Scanner and its Trend (in Japanese). Japanese Journal of Forest Planning 41(1), 1-12.

• Hirata, Y., Furuya, N., Suzuki, M., Yamamoto, H., 2008. Estimation of stand attributes in Cryptomeria japonica and Chamaecyparis obtusa stands from single tree detection using small-footprint airborne LiDAR data. Journal of Forest Planning 13, 303-309.

• Hirata, Y., Furuya, N., Suzuki, M., Yamamoto, H., 2009. Airborne laser scanning in forest management: individual tree identification and laser pulse penetration in a stand with different levels of thinning. Forest Ecology and Management 258(5), 752-760.

Thank you for your attention!


Application of Airborne LiDAR Measurement to Forestryarchive.sokugikyo.or.jp/pdf/gift/2laser/02.pdfApplication of Airborne LiDAR Measurement to Forestry 산림분야항공기LiDAR

Documents