Frontiers in Fuels Science:
Species-Specific Crown Profiles Models from
Terrestrial Laser Scanning
Background
Key Concepts:
 Document accuracy and validity of existing crown biomass equations
(Affleck)
 Develop new crown biomass/fuel equations for inland northwest tree
species (Affleck)
 Parameterize crown fuel density profiles over individual stems per species
with TLS
• Stem dimensions
• Crown characteristics (e.g. shape and density)
Where We’ve Come From:
 Branch Scale Biomass Using TLS
Background: Working Hypotheses (TLS)
 Boles
• Incremental bole diameters used to generate basal area
• Diameters could be used predict biomass from revised allometries
 Crown profiles differ between species
• A large body of previous work on crown shapes
• Laser gives precise measurements and characterize crown
• High replication using laser
• Tests using limited sample sizes suggest between species differences
are more pronounced than those within species
 Biomass is distributed unequally throughout a crown:
• Both vertically and horizontally
• Distribution differs between species
Introduction: Sampling
 Connection to Destructive
Samples
• Collect TLS data of
sampled trees
• Collect independent
samples concurrently
representing:
− same species
− site conditions
Vicinity Map
 TLS Sampling Objective
• n > 200
Trees sampled
Trees & TLS sampled
Introduction: TLS Instrument
Optech Ilris 36D HD
Terrestrial Lidar System
• 10 KHz sampling
• Point density 1cm or less
• Local characterization
Introduction: TLS Sampling
Current Thinking:
Zone of
Occlusion
 Literature suggests that estimating
biomass from TLS requires:
• Scanning from multiple angles
• High resolution
• Offsetting potential occlusion of
data within tree
Project Approach:
 This project seeks to optimize TLS
collection by:
• Limit scanning time by only sampling a
hemisphere
• Control errors of omission through
sampling volume
Zone of
Interception
15m
Minimum
Distance
Introduction: TLS Sampling
Data Resolution a Function of
Range:
 The instrument is parameterized to
collect data at a set resolution at a
determined range (focal plane)
 Characterize hull of tree
Focal Plane
(~4.0mm density)
 Data decreases in density as it
gains range
Types of Occlusion:
 Penetration of energy through the
canopy or objects
• Angle independent/dependent
 Shadowing of canopy elements
• Depends canopy density
• Angularity (branches shadowing
objects above)
Origin
Methods: TLS Collection and Processing
Data Process Flow
 Data Collection
• Largest time commitment (e.g. travel, set up, Etc.)
• ~15 minutes per scan
• Dependent on site conditions (e.g. adjacent tree
density)
 Alignment of Scans (Polyworks)
• Potentially time consuming
 Process Using Lab Developed Applications
• Designed to begin optimizing tree processing for
efficiency and repeatability
• Single processing flow for applying alignment,
calculating bole dimensions, and normalized
canopy distance from
bole.
Methods: TLS-Based Bole Measurements
Three Estimates of
Bole Diameter Up The
Tree:
1.
2.
3.
Three samples of distance
from bole centroid
Fitted line from selected
bole points
Modeled from the initial
bole diameter at the
bottom of the tree
Methods: Distance and Tree Cleaning
Height (m)
Methods: Building a Library of Crown Shapes
Distance (m)
Analysis
• Species used in preliminary work
• Crown Characterization
• Crown Shapes (profiles and lengths)
• Biomass Distribution
• Bole characterization
• Integrating it all
• What’s next?
Analysis
• Data sample for preliminary analysis
• 6 Douglas firs (Pseudosuga mensiesii)
• 3 grand firs (Abies grandis)
• 1 ponderosa pine (Pinus ponderosa)
• 1 western larch (Larix occidenatlis)
http://www.idahoforests.org
Analysis: Crown Profiles
• 90th crown width percentile
chosen to define outer hull
• Points at each height interval
Analysis: Crown Profiles
• Rescaled
both axes
as 0-1
• Did this for
6 Douglas
firs and 3
grand firs
Analysis: Crown Profiles
• Combined
all samples
per species
into one
“uber-tree”
each
Analysis: Crown Profiles
Model Fitting
• curve smoothing
• scale of variability
• exclusion of
bole/incorporation of
crown base height
Analysis: Crown Profiles
Douglas Fir (n=6)
Grand Fir (n=3)
Ponderosa Pine (n=1)
Western Larch (n=1)
Analysis: Crown Base Height
Using some
impartial metric to
consistently define
lower bound of
crown length
Analysis: Crown Base Height
Analysis: Crown Base Height / Crown Profiles
Analysis: Crown Biomass Distribution
• Hull / void
• Survivability analysis
• Hull delineation
• Within-hull biomass
distribution
Analysis: Boles
• Many different radius measures generated
curve
fitted
constant
dist1
dist2
dist3
average
dist
Analysis: Boles
•
Potential use in linking
TLS data to allometry for
biomass prediction
•
Problems to overcome
Analysis: Integrating It All

•
•
•
•
•
Per Species
Apply crown base metric
Generate the “uber-tree”
Fit crown profile curve
Determine hull-void
demarcation
Determine biomass
allocation pattern
 For New Trees
• Need species, DBH,
height, crown length
• Use DBH to calculate
biomass
• Use crown profile
function to build outer
hull shape
• Allocate biomass within
defined hull
Analysis: Where To Next…
Things to think about:
• Occlusions (of bole, inner vegetation)
• Best metric for CBH delineation
• Appropriate scale of variation for crown profile
curve
• Defining hull/void demarcation
• Distributing biomass within that hull
Next (this summer through Spring 2013)
• More trees, more scanning, more data processing
• Linkages to Affleck lab measures
• Exploration of applications beyond fire
The Laser Team:
 Eric Rowell, Ph.D. Plot scale surface fuels
characterization; integration of airborne and
terrestrial scanning, fuel consumption.
 Tara Umphries, M.S. Quantifying fuel dimensions in a
grassland.
 Jena Ferrarese, M.S. Measuring conifer crown
dimensions and the distribution of biomass within
them.
 Theodore Adams, M.S. Defining/distributing fuel
elements in diffuse shrubs of sagebrush and chamise.
Acknowledgements:




Joint Fire Sciences Program
Inland Northwest Growth and Yield Cooperative
Affleck lab
Active Remote Sensing Lab, National Center for Landscape Fire Analysis