Using BigBAF Sampling in a New England Mixedwood
by Kenneth M. Desmarais
Fox Research Forest NH Division of Forests and Lands
Forest Management Bureau
New England forest stands are often highly
variable and local foresters have the difficult problem of sampling these
forests. Traditional sampling methods such as Horizontal Point Sampling (HPS),
can be extremely time consuming if a forester intends to get a point
estimate with reasonable confidence limits. It is not unusual for the
confidence limits by species to be ± 300% or more for a particular
estimate, especially sawtimber or pulpwood volume. The problem is compounded
in the region by local foresters’ preference for the 10 Basal Area Factor
(BAF) prism, which usually requires them to measure about 5 to 20 trees per
point. So for each single basal area observation, 5 to 20 trees must be
measured, based on the stocking of the stand.
An alternative to HPS is BigBAF sampling. This
method uses 2 different BAFs; a large BAF often ranging from 80 to 110
(English) for determining which trees to measure, and a smaller BAF often 10
to 20 (English), for collecting tree counts (basal areas). The large BAF is
very efficient because it dramatically reduces the number of “In-trees”
to measure and most of the “In-trees” are very close to point center.
The smaller BAF permits the collection of tree counts with a much lower
variance than would be achieved with the larger BAF.
Methods and Materials
To test this method, I used BigBAF sampling in a
marked timber sale prior to cutting. The marked trees had received total
enumeration so I had excellent stand volume estimates. Next, I collected
data from 80 points in this 9.1 acre stand. I used an 80 BAF homemade angle
gauge to determine measure trees and a 20 BAF prism for basal area counts.
To measure tree diameter I used a commercially manufactured diameter tape
and ocularly estimated merchantable heights to the nearest ˝ log (8 feet).
The fieldwork required 4 ˝ hours to complete. Calculations were done on a
Microsoft Excel spreadsheet. Point volume estimates are calculated the same
way as in HPS by multiplying the mean VBAR by the mean basal area.
From my sampling, Eastern white pine sawtimber
had a mean Volume to Basal Area Ratio (VBAR) of 192.86 board feet per square
foot of basal area and a mean basal area of 23.25 square feet per acre
resulting in an estimate of 4,484 board feet per acre. The marked pine
volume was 4,908 board feet per acre, a difference of 8.4%. Eastern hemlock
had a mean VBAR of 112.84 board feet per square foot of basal area and a
mean basal area of 25.0 square feet. The hemlock cruise estimate was 2,821
board feet per acre compared to the marked volume of 3,017, a difference of
6.5%. Finally, Northern red oak had a mean VBAR of 97.3 board feet per
square foot of basal area and a mean basal area of 14.75 square feet. The
sawtimber volume estimate was 1,435 board feet per acre compared to the
marked volume of 1,327 board feet per acre, a difference of 8.1%.
I used Bruce’s equation to calculate the
confidence limits of the cruise:
SE%s are calculated by multiplying each standard error of the mean expressed
as a percent, by a “t” value to get a range of values with a certain
degree of confidence. I used a 95% degree of confidence for this cruise and
my SE%s are adjusted accordingly. Each SE% within the square root sign
(radicand) must be adjusted individually because they usually have different
degrees of freedom.
Inserting my white pine values into the equation
resulted in a confidence limit of:
Hemlock had a VBAR SE% of 11.2% and a basal area SE% of 23.9% resulting
in a total SE% of 26.3%. Red oak had a VBAR SE% of 8.7% and a basal area SE%
of 40.5% resulting in a total SE% of 41.4%.
Inspection of the above equation shows that the
higher of the two radicands dominates the equation. Reducing the lesser
radicand has little effect on the outcome. For example, if I were able to
reduce the VBAR SE% for white pine by half, the confidence limits would be
only reduced to 28.8%. However, if I reduce the basal area SE% by half, the
confidence limits would be reduced to 16%. Thus, it is suggested that the
two radicands be balanced for maximum efficiency. If a confidence limit of
±15% were desired, each radicand would need to be 10.6%.
This cruise had VBAR radicands of 7.2%, 11.2% and
8.7 % respectively for white pine, hemlock and red oak. However the basal
area radicands were 28.6%, 23.9% and 40.5% respectively. These statistics
suggest that I would have been better off using a larger BigBAF of possibly
100 BAF and spent a larger proportion of time sampling basal area instead of
Traditional HPS techniques would have caused
increased over-sampling of individual trees and under-sampling of basal
area. I think BigBAF sampling will work well in many New England forest
Thanks to Dr. Kim Iles for developing the BigBAF
method and helping me get started, Dr. John Bell, Dr. Thomas Burk, Dr. Mark
Ducey and Dr. Jeff Gove for their additional assistance and support. The
Caroline A. Fox Fund and the New Hampshire Division of Forests and Lands –
Forest Management Bureau funded this project.