More than Just Lines on a Map: Best Practices for U.S Bike Routes
Relationships between forest structure measurements
1. Relationships between forest
1
structure measurements
Peter Scarth
John Armston
Tim Danaher
Rob Hasset
John Carter
Statewide Landcover and Trees Study
2. Outline
C a n o p y S tru c tu re
F ie ld S it e s &
S a m p lin g
C lu m p in g L e a f A n g le
C lu m p in g M o d e l LA I ↔ BA M odel
FPC ↔ CCP M odel
Conclusions
3. Primary canopy structural measurements
Stand structure → 2 components
Structural complexity
Structural attributes
Structural attributes
Projected Foliage Cover
Projected Canopy Cover
Basal Area
Foliage clumping - Ω
Leaf Area Index
•LAI
•LAIe → Leaf Angle & Ω
4. Measuring Canopies - Where
139° 142° 145° 148° 151°
154°
18° 18°
X
X
X
X #
Charters Towers
#
Mount Isa
21° 21°
X
X X
X
X Rockhampton
#
#
Longreach
24° 24°
X
X X
X
X X
XX
X
X
Charleville X
X
X XX
X X
X
#
27°
X
X
X 27°
X
X #
X
X
Brisbane
154°
139° 142° 145° 148° 151°
11. Measuring Clumping (Ω)
L a rge C a n o p y G a p C o n tin u o u s C a n o p y L a y e r G a p in te r s p e r s e d w ith s m a ll c a n o p ie s
1800
1500
Trac
Downwelling PAR (mMol/s/m2)
1200
Measures
canopy “gaps” 900
Compares 600
distribution to
300
“Random”
Difference is 0
33 35 37 39 41 43 45 47 49
clumping Ω Distance along transect (m)
12. Fitting Trac Omega (Ω) ↔ Clumping Model
1
Omega ↔ CCP
Trac Clumping
modelled clumping
0.95
0.9 There is a fit but…
Omega is a function
0.85
of Zenith!
0.88 0.9 0.92 0.94 0.96
Modelled Clumping
Correct for 1.02
Clumping Correction
zenith
Zenith ↔ residuals 1.00
•Try 1st to 4th order fits
•3rd is best (⇓ error σ2) 0.98
•Agrees with published
work (Chen) 0 12 24 36
Zenith Angle
48 60
13. Improved Omega (Ω) ↔ Clumping Model
Same Relationship but improved confidence
But limited number & variety of sites
1
0.95
Trac Clumping
0.9
0.85
0.86 0.88 0.9 0.92 0.94 0.96
Modelled Clumping
14. Clumping ↔ CCP Models
Can recover within canopy clumping (Ψ)
Possible clumping > 1 implies “Regularity” (overdispersion)
Clumping Factors
1
0.95
Global (Ω)
0.9
Clumping
0.85
Crown (Ψ)
0.8
Ψ eb (1−ccp)
0.75
Ψ = Ω = Ωtrac
Ω a ccp
0.7
0 0.2 0.4 0.6 0.8 1
CCP
15. LAI ↔ Basal Area
Theoretical models → more leaf ≈ more BA
Log(LAI) = a + b log(BA)
•As reported in literature
LAI = a BAb
•b ⇒ 1 4
∴LAI = a BA
•Significant 3
Effective LAI (LΩ)
outliers
2
1
log(1 −fpc)
LΩ=−
0 G →0.5
0 10 20 30 40
Basal Area
16. Leaf Angle Distribution ↔ Basal Area
Estimate G from Trac & Field Data
Pushing data to its limits
Poor fit but sensible result
0.8
Leaf Angle Distribution
50o
0.6
70o
0.4
57o
fpc
log 1 −
ccp
G =− 0.2
LΨ
0 10 20 30 40
Basal Area
17. Leaf Angle Corrected LAI ↔ Basal Area
Slightly different estimator
Improved confidence
High BA site has “flatter” leaves
log(1 − fpc)
4 LΩ = −
a + b BA
Effective LAI (LΩ)
3
2
1
0
0 10 20 30 40
Basal Area
18. FPC ↔ Basal Area Nonlinear Fit
Final model has simple form
But still has wide confidence intervals on Statewide data
1
BA
−
0.8
fpc = 1 − e a + b BA
0.6
FPC
0.4
CCP
100%
0.2
60%
20%
0 10 20 30 40 50
Basal Area
19. FPC ↔ LAI Final Relationship
We have estimated G and Ω
LAI ≈ 1.7 as FPC ⇒ 0 ∴ Mature trees have leaves
fpc = 1 − e − GΩL
8
Leaf Area Index
Steeper Leaf Angles ≈ 70o
6
Within
Canopy
FPC≈ 50%
4
Fitted Leaf Angles ≈ 70o to 50o
2
Shallower Leaf Angles ≈ 50o
0.2 0.4 0.6 0.8
FPC
20. Conclusions & Future Work
Canopy parameters are interrelated
Probably different for regrowth – Sample?
No single “best” structure measure
Leaf scale parameters important
LAI & Leaf Angle
Clumping
Thanks to:
Rob Hasset
Site selection & photography
Sel Counter-
Field campaign
Slatters
Additional field sites
Editor's Notes
Mature Forest Types Time taken Transect Trac Issues in other areas? When? Trac Optimal Time FPC / Leaf angle short term variability
Clumping Define Demonstrate Neyman model development Fitting Clumping Factor
Clumping Define Demonstrate Neyman model development Fitting Clumping Factor
Final function fit Rootfinding and stability Confidence Intervals MontiCarlo Credible Intervals Where wetter/dryer points move
Describe instrument
Angle dependence How derived from field and trac data Why BA used Fit Statistics