Modeling mania    Robert Leverett
   Nov 14, 2005 10:02 PST 

Ed, Will, Jess, John, et al:

    After modeling the Thoreau pine to a satisfying 853 cubes and
remodeling the Grandfather pine to a surprising 1093 cubes on Saturday
and then adding 3 small pines from Mt Tom Reservation on Sunday, the
number of trees in the sample now stands at 49. If the weather gods
smile on us and our colds don't get the best of us Monica and I will
head to Cook Forest State Park on Friday and spend Saturday with Dale
modeling the huge Cook, Seneca, and Longfellow pines. Adding these 3
great PA trees and perhaps a couple of small, young ones there and then
adding 5 or 6 more the following weekend will test the impact of sample
size increase on the regression coefficients. who knows, maybe we can
even derive a useful predictive tool. However, I'm not keeping my
fingers crossed on that one yet. We'll probably need at least 100 trees
before the regression model starts to yield useful results, remembering
that the current regression model predicts a negative volume for the
smallest trees. Not much utility there.

     In the dance of the coefficients, what we are witnessing is the
enormous variation in trunk form that is associated with the white pine.
That doesn't come as news. Independent off fancy measuring gadgets, even
an untrained eye can see endless variations in form, but it is still
tempting for us, the faithful, to think (or hope) that there is a magic
volume formula - a one size fits all.

    One of the objectives of the modeling project has been to get a
better understanding on how volume is added over time. So as an exercise
in the comparison of ratios, I decided to examine the largest and
smallest trees in the modeling sample. The following table presents some
interesting ratios.

Comparison of
Grandfather Tree to Mt Tom #7
(Comparison is Grandfather/Mt Tom #7)

Ratio Value

Age 3.6
DBH 3.3
Height 1.5
Volume 18.2

   These ratios are not meant to suggest that if the dimensions of the
Grandfather tree could be extrapolated back to when it was the age of
the Mt Tom tree, that the Grandfather's dimensions would have been close
to those of the Mt Tom tree. Odds are that it was larger, probably
between 90 and 100 cubes. Assuming 100, the volume ratio from then to
now would be 10.9 to 1.

   At this point, enough modelings haven't been done to warrant in
sweeping conclusions. In addition, other participants including Will
Blozan, John Eichholz, and Jess Riddle need to weigh in. But there are
shadows on the horizon. It will come as no surprise that there is
enormous variability in how Pinus strobus adds volume. I would be
immediately suspicious of any modeling process that didn't divide a
trunk up into many segments. I would be even more suspicious of any
simplified regression model for calculating volume.

   I recall a friend of mine Dr Alan Gordon silviculturist and spruce
geneticist who was frustrated at the whiz kids who sat in front of a
computer and created exotic mathematical models for volumes of spruce
trees, based I guess, on silvicultural data. Alan had very little faith
in those models, although they were ostensibly sensitive of every
variation in the data. Says who? Alan would say.

   What the modeling data shows so far is that there are trends, but the
old giants are very individualistic. They make their own rules. As
spunky young pines, they may have been influenced by the herd instinct.
Their behavior was predictable. But as they gained status, they struck
out on their own.   

Bob.


Robert T. Leverett
Cofounder, Eastern Native Tree Society