Tightly Packed Tree Groups   John Eichholz
  Mar 29, 2004 20:48 PST 

Dale, Bob, Gary,:

I like the idea of studying trees in relations or groups.   Many of our
questions have to do with the relationship of the tree to the site
conditions. General terrain conditions, such as aspect and slope,
proximity to streams, latitude, and overall age or history are common to
the stand. GIS mapping has been presented as a way to quantify this.
The threshold of GPS or survey for every tree may be too high a barrier
if we want to have widespread sampling. But a stand can almost always
be assigned a location, just from the map if need be.

A single cove or terrace, a hillside, or any reasonably well defined
small area contains the trees in the group. The area itself should be
easy to locate, perhaps by GPS on a prominent feature. Written notes
describing the lay of the land and the relation of the trees to each
other and to landmarks, combined with identifying features might be as
useful as GPS to locate individual trees. Tagging and GPS can take place
if desired to identify specific trees. The stand moniker can also serve
to mask the ID of a specific tree when needed, while still giving
information about the site.

What does this mean for data collection? While we continue to collect
data for individual trees, we would also collect data on the grove as a
whole. Trees and groves both become individuals. They have
relationships to each other tree to tree, tree to grove and grove to
grove. The levels of tagging and measuring, grouping in stands, and
locating stands might be a workable, efficient way of associating trees
with site conditions.

Also, a tree can belong to more than one stand. For instance, "Zoar gap
cove, flat area at base of rocks" and "Clark ridge" are both valid stand
names for the 120'+ black cherry trees I measured. Data-wise, a stand
is a collection of trees, (rather than, a tree is in a stand.) Stand
measures can be associated with site attributes, and vice versa. Kind
of abstract, I guess. What do you think?


BTW by my reckoning a thousandth of a minute (the 1 and the 6 on 42
3.071N x 80 11.756W respectively) is only 5 or 6 feet. That must be a
pretty tight stand of maples? Those coordinates lie in a 25' square!

John




  From: Robert Leverett
Sent: Monday, March 29, 2004 2:34 PM

Dale:

   One lesson that I've learned the hard way is that in tightly packed
pine stands, if one tree is above 150, there's a good chance that
several are. In the Trees of Peace, several trees made it into the club
this past growing season, but I had thought they were around 3 feet
shorter and so I was disinclined to put the extra energy into sorting
out their crowns for a couple more years. However, Lisa Bozzuto and
Susan Benoit's fine work of tagging the trees for our long term growth
study for DCR necessitated measuring the tagged trees and that necessity
led to Howard and I going after the residuals in the Trees of Peace
Grove. Well, guess what we found?

   On Sunday, the extra 150s in the Encampment Pines came as a fine
surprise although one reason we went there was to search for another
150. The two new 150s in the ENTS Grove of the Encampment Pines are on
the up-hill side of the trail and are mixed in with trees in the low and
mid-140s. It was sheer luck that John Knuerr spotted the crown of the
Lee Frelick Pine through a peep hole. John immediately recognized its
potential. He did a quick measurement and was in the 157-foot range
class. That illicited an ape call from yours truly. When we took careful
measurements and adjusted for the base, we were in the 155 - 156-foot
class. A new spot from high ground confirmed at least 156.1. It could we
have a higher top. We'll try to establish that this coming Sunday. Could
there be a fourth 150 in the Grove. Yes, there could.

    As mentioned previously, the other 150 is up for grabs as to its
name. The Tree Amigos have put it out to the ENTS list for a vote. I
prefer not to be a candidate. I get enough attention as it is.

    Next Sunday, I'm hoping we can spend another day in the ENTS Grove
identifying all the candidates for naming. Trees above 140 are all
available. I think there are between 10 and 20.

    So presently, ENTS president Will Blozan and vice president Lee
Frelich have trees named for them in the ENTS Grove. Will's tree is the
larger in girth. Lee's tree is the taller. Hmmm.   

Bob

RE: What's the number?   Gary A. Beluzo
  Mar 30, 2004 06:31 PST 

Hi John,

The method of mapping trees that keeps coming up in my literature search
is the INTERPOINT method developed by Boose, Boose, and Lezberg in the
April 1998 issue of Ecology. The method is a simplified, and improved,
technique based on a 1978 Rohlf and Archie paper. If you are any other
one on this list wants me to send the INTERPOINT files let me know. I
think it would be a good protocol for ENTS to use, would allow for us
all to use a standardized method, and it would facilitate the movement
of the data into a GIS database like ArcView or ARCGIS easily.

Here is a basic overview of the method (from the Ecology paper):

1. Establish three (or more) bench marks on the periphery of the stand
such that one is on a corner (#1), a second is say 15m down one side,
and the third is 20m down the other side from #1. Spend the time to get
accurate GPS coordinates of these permanent benchmarks since these will
be the primary references for all other measurements in the stand.

2. Each tree to be mapped is labelled and measured for dbh to the
nearest 0.1 cm. Aluminum tags at dbh height (1.37M) are convenient
reference for setting the height of the measuring tape for each
tree-to-tree measurement.

3. Beginning in the vicinity of the bench marks, and moving gradually
across the plot, each target tree (called subject trees by Susan Benoit
at MTSF) is located by measuring the horizontal distance (may need a
plumb bob) to the three benchmarks (or later three other REFERENCED
trees). The three points should be more than 1M apart; pairs of
reference trees that subtend an angle of <20 degrees or >160 degrees at
the target tree should be avoided. Distances are measured from bark to
bark to the nearest 0.01M. After each tree has been measured to the
three reference points (initially the three benchmarks but deeper in the
stand other reference trees that have already been measured)the target
tree should be tagged so that you can identify it as a potential
reference tree.

4. At the end of the field day the measurements are entered into a
computer, plotted and the error analyed using the INTERPOINT programs.
The resulting plot can then be imported into a GIS database.

All of the details can be found in the Ecology paper.

I also have loaded a new extension to ARCVIEW 3.2 on my laptop called
AVCAD which allows me to plot trees based on combination of angle and
distance measurements (at home or real time in the field!), so I am
going to plot up Susan Benoit's data sometime this week.

I would love to work with folks to get the whole process standardized
and efficient.

Gary

Gary A. Beluzo
Professor of Environmental Science
Holyoke Community College
303 Homestead Avenue
Holyoke, MA 01040
Re: What's the number?   John Eichholz
  Mar 30, 2004 10:44 PST 

Hi Gary,

Thanks for the very detailed reference. I would be interested in
reading the original papers.

I think your general approach can be very useful both at the tree level
and at the stand level. I am still thinking about mapping an object
called a "grove" on the map, using your method or direct GPS to a
monument within the grove. Each grove has its own characteristics, for
example, a Rucker index or a basal area per acre. Groves stand in
relation to each other in space as well. Perhaps a minimum definition
of a grove would be the three required initial benchmarks, with the one
benchmark chosen as the grove location.

How well would the process work at +/- 1 yard distance precision?

John

RE: What's the number?   Gary Beluzo
  Mar 31, 2004 10:14 PST 

Dale,

You need to use GPS to establish accurately the coordinates of your
three initial benchmarks (on the periphery of the stand where you CAN
get GPS measurements) and then once target trees have been accurately
located with the method those trees can and do become reference trees as
you map further into the stand. The important thing is that you begin
with three properly located benchmarks (can be trees as long as you can
determine the coordinates of the center...the method assumes a dbh of
"0" for the three benchmarks).

Gary

Gary A. Beluzo
Professor of Environmental Science
Holyoke Community College
Re: What's the number?   John Eichholz
  Apr 05, 2004 20:39 PDT 

Gary, Bob:

(long) Summary: I have an idea for a mapping protocol that uses at
least three GPS located monuments and a rangefinder and clinometer for
measurements. The resulting data can be loaded into interpoint, and
formatted GIS data obtained as output. This output should be accurate
to +/- a yard or so under field conditions.

The motivation for this protocol comes from your idea of locating some
of the MTSF trees for the purpose of placing them on a GIS layer, which
could open up a world of data to correlate with each tree. The
Interpoint reference was very helpful. I found the complete text of the
article on the Harvard Forest website, along with a downloadable copy of
the INTERPOINT software. I was surprised that it was only about 50k in
size, including the executable, a readme file and a set of test data for
1000 trees.

I wanted to come up with a simpler protocol that would use a rangefinder
and clinometer instead of tapes, to accomodate the often widely spaced
trees in the areas we study. This protocol could be easily tested with
a small data set generated for several trees in a day. The three or more
benchmarks would be set up within view of the subject trees, and at
least 20 yards distant. Then we would take GPS readings at the
benchmarks. The cartesian coordinates (in yards) of the benchmarks
(needed as input) can conveniently be figured out using the
triangulation program provided with INTERPOINT. The distance from each
benchmark to each subject tree is measured using a rangefinder to the
nearest yard, and corrected for slope using the clinometer and the
cosine formula. (Cosine of the angle times the distance equals the
horizontal distance.) The circumference is measured in feet then
converted to yards (same units as distance). There is then enough data
to set up an input file for INTERPOINT. You get the formatted output
from INTERPOINT, and there you go.

When I ran a subset of the INTERPOINT test data (the first 66 rows)
unmodified, the results came within 7cm of true for every tree. I then
rounded the location data from 3 digits to 2, and was able to see the
effect on cumulative error (not much). The trees were then located
within 13cm of true. This data is similar to what you get using the
protocol as you described it.

I modeled the simplified process using graph paper and a ruler. I
established three benchmarks along the edge, and randomly located 8 dots
within. I invented cbh for these trees in decimal feet, then converted
to cm. Scaling up the measurements to about a 20-90 meter spacing from
the benchmarks, and rounding to whole meters, I set up a data file and
ran INTERPOINT. The results came within 58cm, with all but 2 of 24
results within 25cm. The key was that all readings were taken from a
benchmark, so there is no more than a single error to add in. I imagine
that the error from chaining measurements from tree to tree like in the
original protocol, would be at least additive for the first few.

If it would seem reasonable for our purposes to have tree locations
within a couple of yards on the GIS map layer, and we could verify with
a ground test, this method might be useful. Once enough GPS monuments
were once set, trees could be added to the data bank using a rangefinder
and clinometer at any time. And, you could do it solo. What do you
think? Would you like to try this process at the ENTS grove, as well?

John
RE: What's the number?   Gary A. Beluzo
  Apr 06, 2004 07:03 PDT 

Hi John,

I am very interested in Terrain Shape Index (TSI) (and Landform Index in
larger areas) and would like to see your thoughts about it..It is a
simple field method that Henry McNab developed published in the early
1990s. Here is the basic method excerpted from the Forest Science, Vol
35, No. 1 article entitled "Terrain Shape Index: Quantifying Effect of
Minor Landforms on Tree Height" pp 91-104. :

In essence the index represents the mean relative difference in
elevation between the center of the plot (i.e. stand, grove, or cluster
of trees) and its boundary. Elevation of the observer, Zo, defines a
fixed horizontal reference plane. The plot radius, R, when projected on
the cone's surface below the observer, defines a sloping elliptical
plane with a boundary ranging between maximum and minimum elevations at
points Z1 and Zn. The mean difference in elevation between the observer
and the plot boundary is:

Zmean= (Z1-Zo) + (Z2-Zo) +...+ (Zn-Zo) /N

For convex topography the sign of Zmean will be negative because the
mean elevation of the plot boundary is less than the elevation of the
observer. Similar logic is used in visualizing a plot situated on
concave topography, except the value of Zmean is positive. A value of
Zmean near "0" would indicate a nearly linear, but not necessarily
level, surface shape, such as might occur on a side slope.

Although the plot side does not enter into determination of Zmean, the
gradient of surface curvature varies, depending on the size of the area
being described. Standardizing the index for plot size yields:

Terrain Shape Index (TSI)= Zmean/ R

Where Zmean= mean elevation of the sample plot boundary
          R= plot radius measured in the units used for elevation

------------------------------------------------------------------------
------------------------------------------------

The TSI is thus equivalent to the mean slope gradient of the plot
boundary as viewed from the plot center, with units of meters change in
elevation per meter of plot radius. The TSI gives an indication of the
amount of "concavity" or "convexity" of the plot center. This can be
determined at a pixel by pixel basis (at whatever radius from each pixel
you want to use)for the entire USGS quad using GIS!

McNab found that in the Southern Applachians, the distribution and
growth of trees was highly correlated with local topography.
Application of the TSI in three even-aged stands of yellow-poplar on
high quality uniform sites shows that it is highly correlated with total
height of the trees in a stand, with r2 ranging from 0.45 to 0.74. In
comparisons amopng the stands, TSI accounted for an average of 51% of
the variation in site index.

SO, it would be interesting to apply the TSI to our groves in MTSF and
other sites to see how much of the superlative height may be explained
by terrain topography (which of course influences variables like
moisture availability). Also, we may be able to use TSI in combination
with other GIS layers to predict where other superlative trees may be
growing.

Gary




Gary A. Beluzo
Professor of Environmental Science
Holyoke Community College
RE: What's the number?   Gary A. Beluzo
  Apr 06, 2004 07:16 PDT 

John,

Yes..yes..yes...sounds wonderful. You are right of course, that we need
a simple method that folks working SOLO can use to document the location
of trees. I like your idea of simplying the field protocol by using our
standard field tools- the clinometer and laser rangefinder. My only
concern is that the laser rangefinder provides measurements to the
nearest yard rather than centimeter and some of the rangefinders do not
provide distances less than 18 yards or so (so what about trees that are
relatively close to the benchmarks or densely packed?)...but if we are
willing to accept these shortcomings with our existing equipment then by
all means go for it.

As soon as I receive the first grove data set (LATLONG in decimal
degrees preferred, Massachusetts State Plane NAD 1983 the best..that's
what MASSGIS uses and what our GIS database is now in) I'll plot it up
and send you a JPEG with some different background layers so you can see
what the ARCGIS and ArcView 3.2 software programs are capable of. Right
now the main limitation for getting this GIS tree database going is..as
always.. the field data; I have access to all of the MASSGIS data layers
which include color digital orthophotos (0.5M resoluton), digitized USGS
quad sheets, bedrock geology, surficial geology, soils, forest type
(very coarse resolution though), major roads, major streams, watersheds,
ACEC, etc. I will use USGS Digital Elevation Model files to create 3D
models of each site over which we can "drape" whatever data layers
(including any tree attributes like height, girth, health, etc) we want.
I can also generate the various derivative terrain layers (slope,
aspect, Terrain Shape Index, Landscape Shape Index)from the DEM files.
Once the trees are plotted we can do many types of population,
neighborhood, and other statistical analyses. The sky is the limit!

Thanks for all of the time and expertise you are putting into
this...once we standardize the field measurement method so that any of
our ENTS folks can document individual trees easily and in a format that
facilitates inclusion into the GIS database, we'll be entering a new
phase of the work. These are exciting times!

BTW, Tuesdays and Saturdays are good field days for me and after I
return from the Smokies trip on April 26 I can get out on many Sundays.
Feel free to call me at home 413 467-1531 or on my cell 413 210-7150.

Gary