Transcript Spatial Analysis of Large Tree Distribution of FIA Plots on the

Spatial Analysis of Large Tree
Distribution of FIA Plots on the
Lassen National Forest
Tom Gaman, East-West Forestry
Associates, Inc
Kevin Casey, USDA-FS R5 Remote
Sensing Lab
Our FIA Investigation
1. Examine the Statistical Value of
Collecting Hectare Tree Data.
2. Test spatial relationships among trees
using the distance and bearing data
to index plot structure.
3. Create plot “cartoons” and relate to
high-resolution imagery, hence to
landscape
1. Examine the Statistical Value of
Collecting Hectare Tree Data.
2002 Lassen National Forest
Densification & FIA Hex Plots
• 132 plots – 83
have HA trees
• 66 had multiple
HA trees >= 32.0”
dbh
1 ac. Vs. 1 ha.!
What do we gain from the extra work?
Methods
A.
B.
Select large trees on annular plot using
the distance and azimuth data.
Create 2 data sets for each plot
1.
All Hectare Trees
2.
Hectare Trees on ¼ acre
annular plots only
C.
Statistical Analysis # Large
Trees. Acre data
(expanded to ha value)
vs. hectare data by
mapped forest “Type”
The Raw Data
Statistical Analysis of Results
Statistical Analysis of Results
Sample size: Hectare
No. of Large Trees
11.8%
All plots
M easured on full Hectare
Mean
3.857143
Standard Error
0.455235
Median
1
Mode
0
Standard Deviation 5.250026
Sample Variance
27.56277
Range
29
Minimum
0
Maximum
29
Sum
513
Count
133
Sample size: Acre
No. of Large Trees
14.0%
All Plots
expanded to per ha. Basis)
Mean
4.143105
Standard Error
0.589473
Median
2.471
Mode
0
Standard Deviation 6.798135
Sample Variance
46.21464
Range
37.065
Minimum
0
Maximum
37.065
Sum
551.033
Count
133
Conclusions.
1. Though SE consistently somewhat less for HA sampling by type, difference
would reduce with larger # plots.
2. Smaller (4 of the ¼ acre) samples may overestimate # large trees/ha
3. Ha Plots offer an excellent tool for more accurately quantifying large trees on
individual plots but of limited value in reducing the error for large samples.
Statistical Analysis of Results
Mean # HA trees/Hectare by type
9
8
7
6
5
Measured on full Hectare
4
3
2
expanded to per ha. Basis)
All plots
EastSide
Pine
Mixed Fir
Mixed
Pine
Lodgepole
Pine
Red Fir
1
0
Mean
Mean
Mean
Mean
Mean
Mean
Conclusions.
1. Though SE consistently somewhat less for HA sampling by type, difference
would reduce with larger # plots.
2. Smaller (4 of the ¼ acre) samples may overestimate # large trees/ha
3. Ha Plots offer an excellent tool for more accurately quantifying large trees on
individual plots but of limited value in reducing the error for large samples.
2. Test spatial relationships among trees
using the distance and bearing data
to index plot structure.
Creating the “Clumping Index”
Methods
• Selected 11 plots…1 randomly for
each forest “Type” per regional type
map.
• Calculate coordinate
locations of each tree
and create a shapefile
for each plot.
0614151
Tested a variety
of methods which
did not work
0606172
Calculate Proximity
• Use Proximity Analysis tool to create
“PROXIMITY POLYGONS”
• Calculate area of each polygon and total plot
0614151
Calculate “Clumping Index”
• Divide total area by number of trees to obtain
AVERAGE polygon which would represent equal
spacing (as in a plantation)
• Divide actual area of each PROXIMITY POLYGON
by AVERAGE thus obtaining an expression of
POLYGON CLUMPING.
• Average the two lowest POLYGON CLUMPING
values to obtain the CLUMPING INDEX.
Clumping Index Values
0614335
Clumping Index Values
Low Value—Most Clumping
High Value—even distribution
0615554
Clumping Value = 1
no clumping!
Conclusions from Clump Index:
• “Indexing” may be a valuable use of FIA for
ecological modeling of variability within the
landscape
• Would be interesting to examine relationships
among snags and smaller trees
• The index could be derived directly from tree
data without all the interim steps.
3. Create plot “cartoons” and relate to
high-resolution imagery, hence to
landscape.
Mapped the Inventory Plots using Distance and
Azimuth Data from Tree (*.TRE) files
All Trees
Large Trees on the Annular Plot only
Trees on the entire Hectare
Conclusions from SVS process:
• Trees can be realistically drawn in their
true locations using SVS.
• GPS coordinates can be used to locate plot
centers on geo-referenced imagery.
• Annular and hectare plots can be
accurately drawn over imagery.
• In dense stands individual trees are very
difficult to identify.
0615367 clump index 0.19 med high
0615039
clump index 0.166
0614715
(clump index = 0.134 medium high)
0606258 (clump index 0.435 medium)
0615554 (clump index 0.95 low)
Conclusions
• Spatial Analysis allows us to extract more
information on individual plots
• FIA data may be valuable in modeling
relationships
• www.forestdata.com
• www.forestdata.com