(1ru~&(&l::.

?~lft!..
In4rfVI r'Cf-t
Bt--)(-/~r
/1.1-(j

An
Electromechanical
Dendrometer
Band
T H. Hall
C.R. Layton
L.D.Oxtoby

•• Environment
Canada
Forestry
Service
Envlronnemenl
Canada
Service
des For Is

VICTORIA, B.C.
BC,X'184, NOVEMBER, 1978
 2

ABSTRACT

An electromechanical dendrometer band,

designed for use in tree physiology studies (Brix,
1978; Crown and Brett ~ ill., 1975), is described
and evaluated. The dendrometer meets its design
objectives of precision (around 0,2 mm diameter
increment), easy installation, ground level reading
and low unit cost ($5 - $10). Laboratory tests esti-
mated instrument precision to be ± 0.25 mm (95%
confidence interval) and ± 0.16 mm (80% confi-
dence interval). Field tests indicated that band
accuracy may be influenced by an initial period of
underestimation, owing to initial slack upon installa-
tion and by an uneven recording of true increment,
because of a tendency of bands to stick.

INTRODUCTION 1932; Daubenmire, 1945) measure changes in stem

or branch radius at a particular point on the cir-
This report describes and evaluates an electro- cumference. These instruments use screws anchored
mechanical dendrometer band that was developed to the xylem some distance beneath the cambium,
for use in tree physiology studies (Brix, 1978) which to establish a fixed reference point, and record
were carried out as part of an ongoing multidisci- changes in distance between this fixed point and a
plinary research project (Crown and Brett tl ~, position on the outer bark which moves as the tree
1975), established in 1970 near Shawnigan Lake, grows.
B.C. The dendrometer was used to obtain upper
stem diameter measurements of more than 300 trees Several types of transducers have been used to
on a weekly basis over three growing seasons from record increments as electrical signals; for instance,
1973 to 1975. a variable differential transformer (Impens and
Schalck, 1965) and various types of potentiometers
Dendrometers have been developed in a (Phipps and Gilbert, 1960; Dobbs, 1969; Woodman,
variety of forms over the last 50 or more years. 1971; Kenerson, 1973). Electrical outputs can always
Existing designs might be categorized according to be read remotely. Dendrographs, for continuous
what they measure (circumferential increment or recording, have been developed by Fritts and Fritts
radial increment) and how they measure (optically. (1955) and Phipps and Gilbert (1960).
mechanically or electrically; continuously or dis-
cretely in time; remotely or not). Advantages and disadvantages of eXisting
designs have been discussed by Borman and Kozlowski
Band dendrometers (Hall, 1944; Liming, (1962) and problems of accuracy have been raised
1957; Dobbs, 1969) measure changes in circum- by Bower and Blocker (1966) and Auchmoody
ference by encircling a stem or branch with a metal (19761. Generally speaking, dial gauge dendrometers
band. Bands are held taut by a spring or a clip and can provide considerably greater precision than
are able to expand to register changes in length as band dendrometers (around 0.02 versus 0.10 mm
the tree grows. Dial gauge dendrorneters (Reineke, diameter increment), the greatest sensitivity stated
 3

being 0.002 mm by Fritts and Fritts' (19551 dendro· tree, holds the band taut and maintains a given
graph. Some believe that dial gauge dendrometers potentiometer setting for a given stem dimension.
may actually measure anomalous meristematic Leads extend from the potentiometer terminals
activity. which they induce by their physical damage down the stem for reading at ground level. The band,
to cambial tissue when installed. Dendrometer bands, which is free to slide through the mounting pads
on the other hand, have been found to underestimate as stem growth occurs, pulls against the restraining
increment for as long as an entire growing season spring and rotates the potentiometer shaft to change
or until enough growth has occurred to eliminate the potentiometer resistance by an amount propor-
the slack in newly installed bands. Whether one wants tional to increment. Changes in potentiometer resis-
to measure radial increment in order to be sensitive tance are measured by differences in scale reading of
to differential growth of individual radii or measure a locally built null-ohmeter. Bands are installed, by
circumferential increment in order not to be bothered one person climbing the tree with a ladder, before
by this differential growth is purely a matter of the start of each growing season and with enough
experimental objectives. band length to accommodate an expected season's
growth.

INSTRUMENT SPECIFICATIONS AND

DESIGN INSTRUMENT PERFORMANCE

Study objectives defined performance speci· Estimate of Precision

fications which made unattractive the selection
and use of an existing dendrometer design. The plans Laboratory tests, as described below, were
to study stem form and phenology of total tree carried out to estimate dendrometer precision. For
growth suggested the need for a band type dendro- these tests, an assembly was built, using a micro-
meter to measure circumferential rather than radial meter dial and parts from a discarded tree ring
increment. Estimates of growth rates involved sug- analyser, that could hold the dendrometer frame
gested a needed precision of around 0.2 mm diameter firmly in place while rotating its potentiometer
increment - a precision at or near the limit of existing shaft by means of a linear pull measurable to 0.02
dendrometer bands. Plans to collect data for more mm. With this assembly, the relation between null-
than 300 trees on a weekly basis would necessitate a ohmeter readings (in arbitrary units) and linear
very low unit cost ($5 to 510 for the budget avail- displacement was investigated. From this investiga-
able). This large number of units vyould, in addition, tion, dendrometer precision was considered to be
require easy installation and ground level reading to dependent upon three factors: calibration, resolu-
achieve a viable field operation. The design would tion and reproducibility_ Each of these factors is
have to accommodate initial stem diameters ranging dealt with in turn below.
from about 5 to 25 on, with seasonal diameter
growths of up to 1 em. From the more than 300 dendrometers that
had been used, 24 were selected at random for
Two years of development· led to the design these tests. By using the assembly above, calibra·
shown in Fig. 1. A nylon frame, which makes con- tion data (change in null~hmeter reading in arbi-
tact with the tree via two mounting pads, houses a trary units/linear displacement in mm) were obtained
50 ohm, ~ turn, rotary, wire-wound potentiometer for each of these 24 potentiometers, Based on linear
whose shaft holds a 4-em-diameter nylon wheel. regression, the potentiometers exhibited a linear
A 2.54-cm-wide by O.2-cm-thick stainless steel response over their full range with a mean calibration
band encircles the tree, inserting through the mount- constant of 9.81 unitslmm, and with a 95% confi-
ing pads and hooked by an S-bend to the rear mount- dence interval on this mean of ± 0.20 units/mm. For
ing pad as shown. A leader wire attached to this band the weekly diameter increments (ranging from 0 to
and a restraining spring attached to the frame connect a maximum of 0.7 mm) encountered throughout
to each other at a pin on the edge of the wheel. the study, this uncenainty in calibration contributes
Spring tension holds the assembly snugly against the negligibly to a measure of instrument precision
such as a root-mean-square error. If, however, the
• By M. Crown, C.P. Brett, L.D. Ox toby and C.R. dendrometers are to be used to measure much larger
Layton of the Pacific Forest Research Centre, increments, for example, seasonal or yearly incre-
Victoria, B.C. ments, they should be calibrated individually.
 4

Spring

Frame _

Wheel
S

Tree

Direction of
band movement
a5 tree grows

o 2 3 4 5 em
I I I I I

Fig. 1. The Electromechanical Dendrometer Band.

 5

The use of a wire-wound potentiometer included twice in a root·mean-square error. Changes

introduces an instrument resolution which is governed in temperature (from 10 0 to 25 0 C) and physical
by the number of potentiometer windings in con- handling were not observed to affect reproducibility
junction with the size of the wheel attached to its significantly compared to this.
shaft. A potentiometer resistance at any instant is
determined by which winding its sliding terminal is The regression of null-ohmeter units on mm
in contact with. Therefore, linear displacements which linear displacement can be used to place confidence
rotate the shaft so as to move the sliding terminal limits on the inverse estimation of mm linear dis-
across a single winding without moving its point of placement or diameter increment from the observed
contact from one winding to another will not be change in null-ohmeter readings (Natrella, 1969),
resolved. This stepwise behavior is demonstrable by For a single diameter increment inversely predicted
observing discrete jumps in null-ohmeter reading in from two null-ohmeter readings, the confidence
response to continuous rotation of the potentiometer limits are ± 0.25 mm 195%1 or ± 0.'6 mm 180%).
shaft. The mean size of this step was estimated
from the change in null·ahmeter scale units for 50 Estimate of Accuracy
steps for each of the 24 potentiometers to be 6.3
units with a 95% confidence interval of ± 0.2 units Field tests were carried out to observe dendro-
or, using the calibration above, 0.65 ± 0.02 mm meter behavior and assess its accuracy in measuring
linear displacement. diameter increments under operational conditions.
Because the bands were installed anew each growing
season, it was essential to check them far under-
This resolution represents the maximum estimation (Bower and Blocker, 1966; Auchmoody,
difference that can be introduced, by this factor,
1976). Therefore, as one field test, for each of the
between an actual linear displacement (however 300 trees that were measured week Iy over each grow-
large or small) and the displacement that would be
ing season with a dendrometer installation, total
inferred using the instrument calibration above. increment over the growing season was also measured
To see this, consider that a vanishingly small linear
with a diameter tape. Whatever precision is asso·
displacement could be recorded as a 0.64 mm dis-
ciated with individual tape readings, collectively,
placement if the sliding terminal were "just ready"
the tape readings should neither overestimate nor
to move its point of contact from one winding to
underestimate. The comparison of dendrometer
the next or, at the other extreme, any displacement
null-ohmeter readings with tape readings for 50 of
up to 0.64 mm could be recorded as a zero dis-
these 300 installations gave a mean dendrometer
placement if the sliding terminal had "just shifted"
calibration of 9.69 units/mm linear displacement
its point of contact. However, since all settings
with a 95% confidence interval of ± 0.53 units/mm.
within the initial step are equally likely starting
A t-test of this data set with the data set obtained
positions, and all settings within the final step are
using the assembly to investigate instrument preci-
equally likely finishing positions, the recorded
sion, showed no significant difference (probability
change in resistance will have an absolute deviation
< 1%) between the two means. This suggests that the
of one step from the change in resistance of an
bands have not underestimated increment over an
equivalent stepless potentiometer. This deviation
entire growing season and, whatever underestimation
will have a rectangular frequency distribution over
they may be subject to for some initial period after
the range ± 1 step. Therefore, the mean absolute
their installation, that underestimation is not large
deviation will be one half the step size and would
enough to significantly lower the total increments
contribute an instrument resolution term of 0.32 mm
that they record for an entire growing season.
linear displacement to a root-mean-square error.
A second field test was carried out, to study
The precIsion of the null-ohmeter was inves- dendrometer behavior on a weekly basis, imme-
tigated by testing it for reproducibility. Being careful diately after installation and performing near the
to position the potentiometer within a step, this estimated level of precision. On each of three trees,
setting was read repeatedly and found to be repro- eight dendrometer bands were installed, four above
ducible to about 1 ±. 1 units, having a mean absolute and four below a vernier band (Brix, 1972) similar
deviation of 0.81 units (0.08 mm) from the mean to that used by Liming (1957). The vernier in each
reading, and a standard deviation of 1.14 units (0.12 case had been previously installed by at least one
mm). Since two null-ohmeter readings are needed to growing season, and is used here as a standard. The
record linear displacement, this factor would be dendrometer bands were installed with differing
 6

degrees of spring tension - the bands on trees num- respect to initial slack. From the analysis of den·
bered 1,2 and 3 being installed with "low", "medium" drometer precision, differences between vernier
and "high" spring tension, respectively. Each set and individual dendrometer readings up to about
of nine bands spanned a stem length of less than 0.3 mm can be accounted for by instrument
50 cm centered about the vernier which was situated behavior - dendrometer step size of 0.64/11 = 0.2
at breast height. All bands within each set are assumed mm plus vernier precision of about 0.1 mm
to measure the same increment (analyses of the data (Liming, 1957). Differences greater than 0.3 mm
did not show any dependence on relative position). probably indicate initial slack. Inspection of
The results of weekly measurements over 4 weeks is Table 1 shows that:
shown in Table 1. Total increment over the 4-week
period was also measured with a diameter tape. in the first week, more "low" tension bands
(tree 1) had initial slack than either "medium"
The data in Table 1 are subject to different or "high" tension bands (namely, 4 obvious
interpretations. The following is offered: and 2 marginal cases versus 1 marginal case
versus 1 obvious and 4 marginal cases, respec-
11 Looking first at overall vernier and dendrometer tively), and
performance over the 4-week measurement period
by comparing their total readings with the tape after the first week, all dendrometer readings
readings indicates an underestimation by the were within 0.3 mm of the vernier reading,
dendrometer bands and the vernier. One inter- indicating that no slack remained after the first
pretation is to suspect the tape readings in Table week.
1, si nee these read ings contrad ict the resu Its of
the other field test, which were based on a far This comparison suggests that bands should be
greater number of tape readings and, although that installed with greater spring tensions than were
test recorded growth only for an entire growing these low tension bands and that bands can be
season, it did not suggest the degree of dendro- installed snugly enough such that, for the bark
meter underestimation implied by the tape read- conditions encountered here, only a nominal
ings in Table 1. However, each tape reading amount of growth is needed to take up initial
entered in Table 1 is an average of two tape read- slack and end the period of underestimation.
ings, one above and one below the set of nine
bands, and each was established by initial and 3) Inspecting Table 1 for variability among den·
finals readings that were made very carefully, drometer readings shows that:
several times and by more than one person.
Alternatively, one can suspect the ability of a variability (expressed as the standard devia-
single band (dendrometer or vernier), however tion of eight band readings per week per tree)
long in place to record accurately a particular is greater in the first week than in any of the
increment. Distinct from the question of an following 3 weeks, and
initial period of underestimation caused by initial
slack at installation is the question of "jerkiness" for the last 3 weeks, when supposedly under-
of response which would be observed if bands estimation is not occurring and variability
were to stick as they attempted to slide through should be due to instrument precision alone,
their collars or mounting pads (Liming, 1957). dendrometer variability observed under opera-
Band underestimation could, therefore, be caused tional conditions is consistent with the earlier
either by slack or sticking, or both. Since stick- analysis of instrument precision.
ing should not be synchronized between installa-
tions, it should be possible to minimize this
effect at least by installing a suitable number of
bands to record a particular increment. This CONCLUSIONS
latter interpretation does suggest that such a
precaution should be taken if the available pre- The electromechanical dendrometer band
cision of the bands is to be realized. discussed in this report was designed to meet the
constraints of precision (O.2·mm diameter increment).
2) Comparing individual dendrometer readings with easy installation, ground level reading and low unit
vernier readings, and considering the vernier as cost (S5 to $10). The analyses in this report and the
standard, shows dendrometer performance with successful use of the dendrometer in obtaining the
 Table 1. Data from second field test to study dendrometer behavior on a weekly basis

Diameter Increment (mm)

Dendrometer

Tree Week Tape Vernier 1 2 3 4 5 6 7 8

1 0.5 -0.21 0.07 0.14 -{).21 0.22 0.23 0.36 0.52 0.14 0.26
2 0.2 0.11 0.35 0.15 0.11 0.29 0.12 0.03 0.36 0.19 0.12
3 0.1 -0.01 -0.04 0.21 0.02 0.13 0.08 0.05 0.20 0.08 0.09
4 0.1 0.27 0.08 0.06 0.16 0.20 0.18 0.10 0.02 0.13 0.08
TOTAL 1.9 0.9 0.16 0.46 0.56 0.08 0.84 0.61 0.54 1.10 0.54
"
2 1 0.3 0.31 0.12 0.15 0.18 0.37 0.45 0.31 0.00 0.24 0.15
2 0.1 0.28 0.16 0.24 0.15 0.13 0.26 0.16 0.30 0.21 0.07
3 0.1 0.12 0.07 0.22 0.17 0.23 0.06 0.23 0.11 0.15 0.07
4 -{).1 0.06 0.06 0.18 0.20 0.10 0.25 0.14 0.15 0.14 0.07
TOTAL 1.2 0.4 0.77 0.41 0.79 0.70 0.83 1.02 0.84 0.56 0.74

3 1 0.8 0.64 0.68 0.55 0.64 -0.06 0.53 0.51 0.49 0.50 0.24
2 0.5 0.37 0.28 0.30 0.32 0.47 0.31 0.35 0.39 0.35 0.06
3 0.4 0.34 0.64 0.30 0.29 0.44 0.60 0.68 0.28 0.45 0.17
4 0.1 0.33 0.12 0.40 0.38 0.86 0.20 0.32 0.39 0.38 0.22
TOTAL 2.2 1.8 1.68 1.72 1.55 1.63 1.71 1.64 1.86 1.55 1.67
 8

needed data indicate that the dendrometer has Fritts, H.C. and E.C. Fritts. 1955. A new dendrograph for
indeed met its design objectives and fulfilled its recording radial changes of a tree. For. Sci. 1,
purpose. 271-276.

Hall, R.C. 1944. A vernier tree·growth band. J. For. 42,

The precision of the dendrometer could have 742-743.
been improved by using a more finely would potentio-
meter and/or a larger diameter wheel in order to Impens, 1.1. and J.M. Schalck. 1965. A very sensitive elec-
decrease its step size, step size being the major factor tric dendrograph for recording radial changes of a
tree. Ecology 46,183·184.
determining instrument precision. However, the
existing design has successfully met its objectives and Kinerson, R.S. 1973. A transducer for investigations of
such modification which might have increased unit diameter growth. For. Sci. 19,230·232.
cost, would still have left the same problems of
accuracy associated with slack and sticking. This Liming, F.G. 1957. Homemade dendrometers. J. For. 55,
575-577 _
behavior, which seems to be characteristic of band
type dendrometers, suggests that these bands should Natrella, M.G. 1969. Characterizing linear relationships
between two variables, 204·249. IN Precision mea-
be preinstalled (in terms of time or amount of surement and calibration - Statistical concepts and
tree growth) to avoid their underestimating incre- procedures, H.H. Ku, ed., Special publication No.
300, Vol. 1, U.S. Dept. of Comm., N.B.S.
ment because of initial slack, and
Phipps, R. L. and G.E. Gilbert. 1960. An electric dendrograph.
be used in sufficient number to enable reporting Ecology 41, 389-390.
of increment as an average of band readings to
Reineke, L.H. 1932. A precision dendrometer. J. For. 30,
guard against inaccuracies owing to sticking of
692-697.
individual bands.
Woodman, J.N. 1971. Measuring tree grQwth with a com-
puterized data acquisition system. Paper, Western
Mensurationists Annual Meeting, Portland, Oregon,
Nov. 30, 1971.
REFERENCES

Auchmoody, L.R. 1976. Accuracy of band dendrometers.

USDA, For. Servo Res. Note NE-221, Northeastern
For. Exp. Sta, 4 p_

Borman, F.H. and T.T. Kozlowski. 1962. Measurements of

tree growth with dial gauge dendrometers and ver-
nier tree ring bands. Ecology 43, 289-294.

Bower, D.R. and W.W. Blocker. 1966. Accuracy of bands Canadian Forestry Service
and tapes for measuring diameter increments.
J. For. 64, 21-23.
Pacific Forest Research Centre
Brix, H. 1972. Nitrogen fertilization and water effects on
photosynthesis and earlywood-Iatewood produc- Victoria, B.C. V8Z 1M5
tion in Douglas·fir. Can. J. For. Res. 2,467478.

Brix, H. 1978. Private Communication. BC-X-184

Crown, M. and C.P. Brett et al. 1975. Fertilization and

thinning effects on a Douglas-fir ecosystem at
Shawnigan Lake: an established report. Canada,
Dept. of Envir., Can. For. Serv., Pac. For. Res. November, 1978
Cen. BC-X-110, 45 p.

Daubenmire, R.F. 1945. An improved type of precision

dendrometer. Ecology 26, 97-98.

Dobbs, R.C. 1969. An electrical device for recording small

fluctuations and accumulated increment of tree
stem circumference. For. Chron. 45, 187·189.