forest research (2010 2019)€¦ · [email protected] technologies perspectives plot origins the...

6/19/2019

1

Professor Chris HopkinsonResearch Chair

[email protected]

TechnologiesPerspectivesPlot originsThe frontierNext steps

Dr. Zhouxin Xi (ABD)Research [email protected]

Hopkinson & Xi, CIF, 2019

TLS Forest Research Trend (2010‐2019)

Web of Science Database

Topics: terrestrial, tree, lidar

Search year: 2010‐2019

Results: 530 publication records

Topic of TLS becomes increasingly popular over the decade


6/19/2019

2

Consumer‐based Flash LiDAR Xbox Kinect

Hopkinson & Xi, 2019

Configurations

• Static frame = HV dual axis scanner

• Rotating head = single vertical line scan

• Combo = frame with pan / tilt mount

• Precision / resolution = mm to cm

• Range = metres to kilometres

• Most combine GPS & image capture

• Some apply motion compensation with IMU

Z+FTrimble

Optech ILRIS Riegl

LeicaEchidna

More typical commercial TLS sensor types


Faro

Teledyne Polaris

6/19/2019

3


Mobile solutions

CSIRO Team:

Zlot, Bosse, Wark, Flick, DuffZeb 1

LiBackpack

Frame scan configuration


6/19/2019

4

At least 3 scans for uniform sampling

Complex set up requires tree height

Occlusion into plot centre

PSP sampling configurations: Frame scan


A + BA B

Frame scanner with pan/tilt mount

OR: rotating line scanner; e.g. Echidna

Occlusions heading out of plot

PSP sampling configurations: Hemisphere scan


6/19/2019

5

PSP sampling configurations: Mobile scan


Truck mountedBackpack mounted

(LiBackpack 50)

https://greenvalleyintl.com/applications/finding‐dbh‐values‐handheld‐mobile‐lidar/

Point cloud attributes: Density & Occlusions

Density

Terrestrial

Airborne

6/19/2019

6

Point cloud attributes: Scanner, Elevation, Intensity, Echo class


Assessing forest metrics with a ground‐based scanning lidarHopkinson, Chasmer, Pow, Treitz. 2004

6/19/2019

7

Tree stem identification & extraction

Neighbourhood

around stem centroid

at max crown size

Point data extracted

written to “tree” file

Tree files “cleaned”

Crown diameter

Manually assessed with virtual calipers

6/19/2019

8

Plot 6 ILRIS tree height

y = 1.09x

R2 = 0.86

0

5

10

15

20

25

30

35

0 5 10 15 20 25 30 35

ILRIS (m)

Ma

nu

al

(m)

1:1 line

Plot 6 ILRIS stem diameter

y = 1.05x

R2 = 0.98

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

ILRIS (m)

Ma

nu

al

(m)

1:1 line

Height

Tree Height Vector connecting max

& min pt within file

Vector length ~ tree ht

Tree stem data from

1.25 – 1.75 m selected

Cylinder “primitive”

fitted

Stem DBH

DBH

Workflow

Plot scans →

Align →

DEM →

Tree isolation →

Stem extraction →

DBH, height, stem taper/volume →

Allometric equations applied

Xi, Z.; Hopkinson, C.; Chasmer, L. Automating Plot‐Level Stem Analysis from Terrestrial Laser Scanning. Forests 2016, 7, 252.

Automated allometric biomass estimation


6/19/2019

9

Workflow

Decompose point cloud into classes

• Stem

• Branch

• Other

Apply biomass factor to each class

Quantitative Structural Modelling (QSM)


Component Maple Aspen Pine

Stem 92.3% 81.2% 91.5%

Branch 6.5% 17.6% 8.4%

Isolated wood 0.3% 0.05% 0.03%

Understory 0.9% 1.2% 0.1%

plotwood

treebranch

stem

understory

isolatedleaf

Xi, Z.; Hopkinson, C.; Chasmer, L. Filtering Stems and Branches from Terrestrial Laser Scanning Point Clouds Using Deep 3‐D Fully Convolutional Networks. Remote Sens. 2018, 10, 1215.

Workflow:

7 TLS plots/582 stems →Machine learning segmentation → Deep learning classification → 92% accuracy

Artificial intelligence species classification


Xi, Z. 2019. Enhancing Forest Biomass Inventory with Ground‐based LiDAR. PhD Thesis, University of Lethbridge. pp162

Lodgepole Pine Red Pine Scots Pine Silver Birch Norway Spruce Trembling Aspen Sugar Maple

6/19/2019

10

Ongoing Research

Hardware

• Multispectral lidar

• Low‐powered real‐time LED/RGB monitoring

• Data fusion & scaling (TLS for cal / val)

Software

• Machine learning, artificial intelligence & deep learning

• New allometric approaches

Applications

• Shrub & juvenile biomass assessment

• Change detection (growth, stress, mortality, seasonal)


Fusion Scanning SystemZhouxin Xi

Ecosystem Diagnostic Imaging Facility

• Western Economic Diversification Project

• Airborne Multispectral Lidar, TLS, UAV

• Environment & Resource Monitoring

• Risk Assessment & Mitigation

• Disaster Response

Thanks!

forest research (2010 2019)€¦ · [email protected] technologies perspectives plot origins the...

Documents