Soggy
but productive weekend, MA |
Robert
Leverett |
Oct
11, 2005 09:43 PDT |
ENTS:
A soggy weekend of
tree modeling had me questioning my sanity.
Nothing new there. I do that regularly. But luckily, most of the
period
was manageable for measuring.
On Friday, Monica
Jakuc and I headed to Bullard Woods in
Stockbridge MA. The objective was to model a huge pine that I
call the
Bullard Woods pine (not very inventive). Its girth 4.5 feet up
from at a
low point on otherwise fairly level terrain is an impressive
13.9 feet.
Its full height is 133 feet. It is very bulky for the first
90-95 feet.
Then it slenders up very quickly. The first modeling with the Rd
1000
incorporated 28 diameter measurements that yielded 845 cubes of
trunk
and limb volume. The trunk portion of the 845 is 761 cubes. On
Monday, I
developed a regression-based model for trunk and limb volume
using data
from the modeling of 24 trees. The independent variables are DBH,
full
height, and diameter at 50 feet. The multiple regression
coefficient
from the sample of 24 is 0.92 - not bloody bad. That's an
R-square of
0.85 for those who prefer that measure of fit. Interestingly,
the model
applied to the Bullard Woods pine yields 880 cubes as opposed to
the
845. That's a difference of only 4%.
In an interesting second
exercise, I treating the trunk of the
Bullard Woods tree as though it were a cone from 4.5 feet up to
its full
133-foot height. The lower 4.5 feet was modeled as a frustum of
neloid
form. This modeling yielded 702 cubes for just trunk versus the
761
obtained from the full modeling. I actually expected the
difference to
be greater because the lower part of the tree is very bulky. At
any
rate, the Bullard Woods pine is large, but falls short of the
volumes
for the biggest of the Great Smoky Mountain hemlocks of
comparable
girth.
On Sunday, I modeled a huge
pine in a Conway MA grave yard. Its
CBH is 14.3 feet and its height is 140.5 feet. It is very large
to a
limb whirl at about 25 feet. The trunk comes down in size
rapidly
thereafter. The modeled trunk and limb volume turned out to be
858
cubes based on 16 diameters. The regression-based model yields
930
cubes. The trunk accounts for 789 cubes of the 858. Treating the
trunk
above 4.5 feet as a perfect cone and neloid below yields 805
cubes. So
we have 858 versus 930 and 789 versus 805. Examining these
results
suggests to me that I need to march back out there and do more
modeling.
What I really need is more measurements from a height of 80 feet
and
over. However, trunk visibility above 80 feet is extremely poor.
This
tree is going to be a challenge in more ways than one. As I walk
around
the graveyard, I try to be mindful and respectful of the
deceased souls.
I frequently stop and nod in respect. As big tree sites,
graveyards need
to be treated with full recognition that other people might see
tree
modeling as deflecting attention from the purpose of the site.
On Monday, Monica Jakuc and I went
to a Mass Audubon property in
Williamsburg MA to model a large charismatic pine that I dubbed
the
Graves Farm pine. I had been told about it by a land
conservationist who
thought the pine fully as impressive as any he'd seen at the
William
Cullen Bryant homestead. The Graves Farm pine is an impressive
11.6 feet
CBH and 130 feet tall. It looks taller, but 130 was all I could
get and
that seems to be in agreement with the growing conditions. The
big pine
is no spring chicken. My best guess is an age of 200 years or
more.
Perhaps 230 or 240. The modeling yielded a surprisingly low 553
cubes of
which 508 are trunk. This result is based on total of 26
diameter
readings. The regression model gives 604 cubes for trunk and
limb
combination versus 553 cubes modeled. The conical model gives
499 cubes
for the trunk alone compared to the 508 for the full model.I
would have
expected a value around 610 cubes. More work is needed.
Fortunately, the
big tree is quite accessible.
The relatively high R value
obtained from the multiple linear
regression suggests that a really gee whiz model could be
developed from
as few as 4 independent variables: height, DBH, diameter at 50
feet, and
diameter at 100 feet. Unfortunately the model could not be
applied to
short trees nor to multi-stemmed trees.
The lesson being
learned (and relearned) from my often erratic
looking data is that a big tree needs to be modeled multiple
times with
comparisons made to other big trees of the same general form.
One cannot
do too much analysis. However, the use of commonsense should
trump
results from applying exotic statistical models. If a number
looks
fishy, chances are that it is, regardless of what kind of exotic
calculation produced it. Nonetheless, I think the regression
model can
point to modelings that need to be revisited. Here is a case in
point.
Tree Full
modeling Regression Trunk
Modeled Neloid-Conical Model Diam
at 80 ft
Mt. Tom HM 758 869
682 919 19.0
DB HM
804 805
748 559 30.6
The Mount Tom hemlock
slenders down rapidly and the Dunbar Brook
hemlock does not. It is a column. I need to remodel both trees.
However,
adjacent measurements for both trees suggests the 19 inches is
reasonable as is the 30.6. Maybe the Mt. Tom tree is a little
low and
the DB hemlock is a little high. Bring the DB hemlock down to 27
inches
and the total volume decreases to 779. Bring the Mt Tom up to 22
and the
volume increases to 744. The result is still in favor of the DB
hemlock.
So many trees to
"remodel". So little time.
Bob
|
|