Tree Physiology 22, 641–648
© 2002 Heron Publishing—Victoria, Canada
Influence of natural temperature gradients on measurements of xylem
sap flow with thermal dissipation probes. 1. Field observations and
possible remedies
F. DO1,2 and A. ROCHETEAU1
1
IRD–CEFE/CNRS, 1919 Rte de Mende, BP5051, 34293 Montpellier, France
2
Author to whom correspondence should be addressed ([email protected])
Received December 14, 2000; accepted September 15, 2001; published online May 1, 2002
Summary The thermal dissipation method is simple and
widely used for measuring sap flow in large stems. As with several other thermal methods, natural temperature gradients are
assumed to be negligible in the sapwood being measured. We
studied the magnitude and variability of natural temperature
gradients in sapwood of Acacia trees growing in the Sahelian
zone of Senegal, analyzed their effects on sap flow measurements, and investigated possible solutions. A new measurement approach employing cyclic heating (45 minutes of
heating and 15 minutes of cooling; 45/15) was also tested.
Three-day measurement sequences that included 1 day without
heating, a second day with continuous heating and a third day
with cyclic heating were recorded during a 6.5-month period
using probes installed at three azimuths in a tree trunk.
Natural temperature gradients between the two probes of
the sensor unit, spaced 8 to 10 cm vertically, were rarely negligible (i.e., < 0.2 °C): they were positive during the night and
negative during the day, with an amplitude ranging from 0.3 to
3.5 °C depending on trunk azimuth, day and season. These
temperature gradients had a direct influence on the signal from
the continuously heated sensors, inducing fluctuations in the
nighttime reference signal. The resulting errors in sap flow estimates can be greater than 100%. Correction protocols have
been proposed in previous studies, but they were unsuitable
because of the high spatial and temporal variability of the natural temperature gradients. We found that a measurement signal derived from a noncontinuous heating system could be an
attractive solution because it appears to be independent of natural temperature gradients. The magnitude and variability of
temperature gradients that we observed were likely exacerbated by the combination of open stand, high solar radiation
and low sap flow rate. However, for all applications of the thermal dissipation method, it is wise to check regularly for natural temperature gradients by switching off the heater.
Keywords: Acacia tortilis, drylands, Granier probe, Sahelian
zone.
Introduction
Thermal methods of measuring sap flow can provide quantitative estimates of whole-tree transpiration or root water uptake,
which are of interest in water balance studies. Although these
techniques are becoming increasingly important in forest science (Smith and Allen 1996, Wullschleger et al. 1998), their
utilization in arid zones such as Sahelian Africa is still relatively rare (Brenner et al. 1991, Allen and Grime 1995, Roupsard et al. 1998).
For large stems, Granier’s (1985, 1987) method, based on a
continuously heated thermal dissipation probe (TDP), is increasingly being used because of its simplicity, low energy requirement, reliability and low cost (Andrade et al. 1998, Braun
and Schmid 1999). However, one drawback of the TDP
method, as with several other thermal methods, is the assumption that, in the absence of intentionally applied heat, the wood
in the section of stem being measured is uniform in temperature. It is known that significant natural temperature gradients
may exist in plants growing in some environments, and several
methods of correcting for their influence on sap flow measurements have been proposed (Èermák and Kuèera 1981, Cabibel
and Do 1991, Goulden and Field 1994, Guttiérrez et al. 1994,
Köstner et al. 1998). The overall reliability of these corrections is mainly dependent on the spatial and temporal variability of the natural temperature gradients, and such information
is scarce in the literature.
Preliminary observations based on use of the TDP method
in an Acacia stand in the northern Sahel revealed large natural
temperature gradients of up to 1.0 °C between the two probe
needles. The aims of this study were to: (1) characterize the
long-term magnitude and variability of natural temperature
gradients in the trunks of Sahelian Acacia trees; (2) analyze
the consequences of these gradients for sap flow measurements; (3) discuss alternatives for correcting for the influence
of temperature gradients on sap flow measurements; and (4)
demonstrate the potential utility of a thermal dissipation signal
derived from noncontinuous heating as a means of minimizing
the influence of natural temperature gradients on sap flow
measurements.
642
DO AND ROCHETEAU
Materials and methods
Granier system
Measurement of sap flow by the method of Granier (1985,
1987) is based on two cylindrical probes, 2 mm in diameter
and 20 mm in length (Umweltanalytische Produkte, Munich,
Germany) inserted radially in the xylem and spaced vertically
by 8 to 10 cm. The upper probe is continuously heated, whereas the lower probe is unheated and the resulting temperature
difference is measured with copper constantan thermocouples
placed in each probe (about 0.040 mV °C –1).
Under zero flow conditions, the temperature around the
heated probe increases to the point where heat dissipation by
conduction through the xylem is in equilibrium with the heat
energy supplied, giving a maximum temperature difference.
When sap is flowing, heat dissipation increases by convection
and the temperature difference decreases. Under laboratory
conditions, Granier (1985) obtained a relationship between
sap flux density (Ju ) and the ratio of temperature differences,
called the flow index (K; dimensionless):
J u = α K β,
(1)
where β = 1.231 and, depending on the units of Ju, α = 119 ×
10 –6 m3 m –2 s –1 or 4.284 l dm –2 h –1. Flow index is defined as:
K = ( ∆T0 − ∆Tu ) / ∆Tu ,
(2)
where ∆T0 is the maximum temperature difference obtained
under zero flow conditions and ∆Tu is measured temperature
difference at a given flux density Ju. To calculate total flow
rate, the conducting sapwood area at the height of the heated
probe is multiplied by mean sap flux density.
Site and experimental design
The study was conducted at Souilene (16°20′ N, 15°25′ W), in
the sandy Ferlo of northern Senegal, 20 km south of Dagana in
the sensu stricto Sahelian ecoclimatic zone (Le Houérou
1989). At Dagana, mean annual rainfall is 283 mm (1918–
1990), mean annual temperature is 28.7 °C and potential
evapotranspiration is 2031 mm year –1. Four main seasons are
recognized in the annual cycle: the rainy season, which lasts
2 or 3 months between July and September; the deferred season, which extends from October to November and is characterized by the drying out of the upper soil layers and the
senescence of annual grasses; the cool dry season, which extends from December to February; and the hot dry season,
which extends from April to June.
The sparse woody community is dominated by three species: Acacia tortilis (Forsk.) Hayne ssp. raddiana (Savi)
Brenan and Balanites aegyptiaca (L.) Del in the tree layer
(90 individuals ha –1), and Boscia senegalensis (Pers.) Lam in
the shrub layer (50 individuals ha –1).
Sap flow was measured in a mid-sized Acacia tree, 7 m tall
and 0.2 m in diameter at breast height, with 30 m2 of crown
projected area. Three probes were inserted at 1.3 m height at
azimuths North (N), South-East (SE) and South-West (SW).
For each probe, two holes were drilled with a vertical spacing
of 80–100 mm. The hole was drilled to a depth of 30 mm to ensure that the sensor was located completely within the active
sapwood area. The conducting sapwood comprised a ring of
wood about 50 mm wide immediately beneath the 5-mm-thick
bark. Mean radius of the heartwood (Rheartwood; delimited by its
red color) was estimated by correlation with over-bark diameter (Dext ) based on observations of fresh sections (Rheartwood =
0.48Dext – 5.02, r 2 = 0.93, n = 14).
After probe insertion, the exposed parts of holes and needles
were coated with silicone. The trunk area containing the
probes was protected from direct solar radiation by a cylindrical deflector 30 cm high. Probes were connected to a data logger (21X, Campbell Scientific, Leicester, U.K.) and signals
were recorded every 5 min. Power circuits, connected to the
heated wire of the upper needles, were equipped with switches
controlled by the data logger.
The experimental program integrated measurements according to a 3-day sequence (see Figure 1). During the first
day, the probes were unheated so that the natural temperature
gradient (dT ) could be recorded. During the second day,
probes were continuously heated according to Granier’s
(1985) original method. On the third day, a cyclic heating system that employed 45 minutes of heating followed by 15 minutes of cooling was tested. The cyclic heating system was
derived empirically to give maximum temperature differences
close to those obtained from probes that were heated continuously (see Figure 1). The new signal derived from the cyclic
heating system was called the alternate signal (∆Ta ) and was
defined as:
∆Ta = ∆Ton − ∆Toff ,
(3)
where ∆Ton is the temperature difference at the end of the heating period and ∆Toff is the temperature difference at the end of
the cooling period.
During a 6.5-month study, from mid-February until the end
of August 1999, 64 3-day sequences were monitored. Several
measurements performed on two additional trees of similar
size suggested that the behavior of each tree was unique. The
tree described here was selected because it showed the greatest
heterogeneity in temperature gradients between azimuths and
thus contained the largest potential for analysis of relationships common to all three azimuths.
At the beginning of the hot dry season (April), leaf fall is
common throughout the Acacia stand, with variable intensity
and duration depending on the tree and the year. In April 1999,
the instrumented tree was at the defoliated stage, a phenological classification defined as more than 90% of the canopy
having no leaves.
Micrometeorology
The local microclimate was monitored in an open field, 50 m
from the tree. A data logger (21X, Campbell Scientific) recorded hourly means of air temperature at 2.0 m measured
with a thermistor (Model 107, Campbell Scientific) and incoming shortwave radiation measured with a silicon pyran-
TREE PHYSIOLOGY VOLUME 22, 2002
SAP FLOW MEASUREMENT WITH THERMAL DISSIPATION PROBES
ometer (LI-200, Li-Cor, Lincoln, NE). Sapwood temperature
of the study tree was measured at the level of the lower needle
of the SW probe with a thermocouple probe (T type, TC S.A.,
Dardilly, France) connected to the data logger used to record
sap flow measurements.
Results
Daily course of natural temperature gradients and their
consequences
Examples of the daily patterns of natural temperature gradients (i.e., Day 1 of the 3-day measurement sequence) are given
in Figure 1a for three probe azimuths on the same trunk.
Global solar radiation, air temperature and sapwood temperature at the height of the SW lower needle are illustrated in Figure 1b. The short time frame involved and the uniformity of
prevailing meteorological conditions support the assumption
of a similar regimen of sap flow and natural temperature gradients over the 3-day cycle. Meteorological conditions were
usually not so uniform over the 64 measurement sequences.
Results from the second day, when heat was applied, show
recordings typical of a TDP system in normal operation: maximum stable temperature differences of about 10 °C during the
night and a decrease in the temperature difference of 2 to 4 °C
during the day, depending on probe azimuth. In theory, decreases in the temperature difference should be wholly linked
to heat dissipation by the convective effect of sap flow. In practice, measurements obtained with unheated probes on Day 1
revealed positive natural temperature gradients from 0.5 to
1.0 °C during the night, and negative gradients from zero to
–2.0 °C during the day. Differences in probe azimuth resulted
in large differences in absolute values and dynamics of natural
temperature gradients, particularly during the day. Among
probe azimuths, the N probe exhibited the smallest natural
temperature gradients.
The positive and negative signs of the night and day natural
temperature gradients led to a twofold overestimation: overestimation of the reference temperature difference at zero flux
and overestimation of the daytime decrease. For example,
75% of the 4 °C decrease observed on the SE probe was induced by natural temperature gradient dynamics.
The magnitude of natural gradients over such short distances (10 cm) revealed that the trunk was a site of important
thermal exchanges. Wood temperature at the lower needle
(SW azimuth), protected from direct radiation, had a pattern
similar to that of air temperature measured at 2.0 m (Figure 1b). In February, at the end of the cold dry season, the daily
range was 15 °C, which increased over 4 h during the morning
and decreased over 12 h during the night. At night, positive
gradients revealed that the top parts of the trunk were warmer
than the lower parts of the trunk. During the day, the situation
was generally reversed but much more heterogeneous depending on probe azimuth, suggesting the influence of more local
factors, such as differences in solar radiation around the trunk
or the effect of low sap flow density combined with temperature of the flowing sap.
643
Based on values in Figure 1, subtracting natural gradients
measured the previous day from values obtained from the continuously heated probes permits an approximate calculation of
sap flux density. Maximum sap flux densities and relative errors (without subtracting natural gradients) were dependent on
probe azimuth: N azimuth (0.8 l dm –2 h –1, +45%), SE azimuth
(0.4 l dm –2 h –1, +288%) and SW azimuth (1.8 l dm –2 h –1,
+86%). Accuracy of these calculations was sensitive to the actual extent of similarity between the natural temperature gradients and the sap flows during the 2 days.
Overestimation by the SE probe was large in relation to the
high natural temperature gradient and low sap flow rate. These
two factors may be linked. The solution coming from the soil
in the morning can be warmer than the sap in the sapwood at
1.3 m, and at the maximum sap flow rate calculated for the SE
probe, sap will take 2.5 h to reach the second needle 10 cm
higher in the stem. This phenomenon can account for strong
negative gradients in the daytime.
The possibilities of natural gradient induction are numerous
and will vary depending on the environment. Our objective
was not to gain an understanding of these phenomena, but to
find a simple and effective way to avoid the measurement errors that they may cause. Hence, it is necessary to know the
long-term variability of the temperature gradients.
Temporal and spatial variability of temperature gradients
Over the 6.5-month recording period, natural temperature gradients were never zero, and they varied markedly depending
on azimuth on the tree trunk, time of day and season (Figure 2a). For the N, SE and SW probes, respectively, the mean
maximum gradients that occurred during the night (dTmax)
were (°C, ± SD) +0.4 ± 0.22, +0.8 ± 0.20 and +1.1 ± 0.36, and
the mean minimum gradients that occurred during the daytime
(dTmin) were –0.4 ± 0.17, –1.5 ± 0.56 and –0.9 ± 0.51.
At the same probe, natural temperature gradients changed
over both short and long time frames, but there were some
short periods of stability (Figure 2a). Between probe azimuths,
there were no overall correlations for natural gradients. However, a seasonal trend of decreasing amplitude of the daily
temperature gradient (dTmax – dTmin) was noticeable with minima in the rainy season (July and August), which is the period
of lowest global solar radiation (Figure 2b). Comparison of
probe azimuth with climatic data showed the greatest correlations with daily air temperature amplitude if the three orientations are separated (data not shown: r 2 = 0.30 (N), 0.42 (SE)
and 0.56 (SW)). The range of daily air temperature amplitude
was wide, with a maximum of 25 °C in the hot dry season
(May) and a minimum of 6 °C in the rainy season (Figure 2b).
The high spatial variability of temperature gradients emphasizes the complexity of gradient induction, and the importance
of localized factors suggests a large heterogeneity of natural
temperature gradients among trees growing in an open stand.
Fluctuation of nighttime continuous TDP signals
In the field, the effect of natural temperature gradients is difficult to detect in the daytime because it is masked by the
dynamics of solar radiation and sap flow rate. At night, the ef-
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
644
DO AND ROCHETEAU
Figure 1. Daily course of thermal dissipation probe (TDP) signals and climatic variables. (a) The TDP signals or
temperature differences for three probe
azimuths on the same trunk: N, SE and
SW; data for the first day were obtained from unheated probes (natural
temperature gradients); data for the
second day were obtained from continuously heated probes (Granier system);
and data for the third day were obtained from probes subjected to cyclic
heating (45 min of heating and 15 min
of cooling). (b) Solar radiation (RS),
air temperature at 2.0 m (Tair), and sapwood temperature at lowermost TDP
needle installed at azimuth SW
(Twood_SW). Recordings are for February 19–21, 1999.
fects are more evident: gradient variability between different
nights should induce high fluctuations in maximum temperature difference (∆TM) at zero flux, which violates the TDP
theory. A progressive variation in ∆TM may be linked to changing conditions of heat dissipation, which may be caused by the
shrinking of wood around the needle after insertion, or seasonal variation in wood water content. A rapid change is more
likely to be a result of the influence of a natural temperature
gradient or possibly a response to nighttime sap flow. As emphasized by Burgess et al. (2000), the TDP system alone can-
TREE PHYSIOLOGY VOLUME 22, 2002
SAP FLOW MEASUREMENT WITH THERMAL DISSIPATION PROBES
645
not determine zero sap flow. However, during April, the
instrumented tree was largely defoliated, so nighttime sap
flows should be close to zero and daytime sap flows should be
very low.
For the continuous TDP system, two nighttime signals were
analyzed: a raw maximum temperature difference (∆TMr), corresponding to the second day in Figure 1, and a corrected
maximum temperature difference (∆TMc), which involved subtraction of the natural gradients recorded during the preceding
day. Because the absolute value of ∆TM at zero flux is dependent on each probe–sapwood system, fluctuations were analyzed by examining the deviations from individual means and
are represented for the three probes simultaneously (Figure 3).
Variation in the raw signals (∆TMr) was large and differed
among probe azimuths, from 0.5 (N) to 1.0 °C (SW), i.e., up to
10% of the mean signal. Variation in the corrected signals
(∆TMc) remained large, from 0.4 (SE) to 0.7 °C (SW), but the
correction markedly narrowed the difference among probe az-
Figure 2. Seasonal course of natural temperature gradients and climatic variables. (a) Daily maximum and minimum values of natural temperature gradients according to three probe azimuths in the trunk (N,
SE and SW). (b) Maximum and minimum air temperatures at 2.0 m
(Tair_min, Tair_max) and solar radiation (RS). Data are plotted at 3-day
intervals corresponding to Day 1 of each of the 64 measurement sequences.
Figure 3. Deviation from means of maximum thermal dissipation
probe (TDP) signals with continuous heating, recorded at night at
three probe azimuths (N, SE and SW) for the raw signals (∆TMr) and
the corrected signals (∆TMc), where natural temperature gradients recorded the preceding day were removed from the raw signal. For the
three probe azimuths (N, SE and SW, respectively), means (± SD)
were (°C): 10.3 ± 0.23, 10.2 ± 0.25 and 10.4 ± 0.49 for the raw signals;
and 10.1 ± 0.25, 9.6 ± 0.21 and 9.7 ± 0.36 for the corrected signals.
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
646
DO AND ROCHETEAU
imuths (Figure 3) and changed the ranking of the absolute
means (from SW > N> SE to N > SW > SE).
However, the correction did not fully stabilize the nighttime
signals, and a seasonal pattern was noticeable (Figure 3). Because it is unlikely that changes in wood water content differed
among azimuths on the same tree trunk, nighttime sap flows
most likely contribute to the variability of the corrected signal.
Based on the defoliation stage of the tree, only the higher values recorded between April and June will be true zero flux signals (Figure 3).
Assuming that the long-term mean of corrected signals
( ∆TMc ) adequately accounts for natural temperature gradients
and that there is a negligible influence of nighttime sap flows,
the difference between ∆TMc and individual raw signals (∆TMr)
should correlate with the maximum natural gradients presented in Figure 2. The results in Figure 4 confirm the hypothesis and the effect of the natural gradients. Data for the three
probes are integrated in a global linear relationship that intercepts zero (r 2 = 0.48).
Noncontinuous TDP signals
As shown in Figure 1, there was evidence that the noncontinuous signal after 15 min of cooling (∆Toff) was linked to the natural temperature gradient (dT ), but ∆Toff did not measure dT
accurately. For the data in Figure 1, the difference between the
course of ∆Toff and the course of dT is expressed in Figure 5
versus corrected sap flow density, calculated with Equation 1
from the continuous signal with gradient subtraction. The values differ from zero and appear to be linked to sap flow rates.
The relationship suggests a pattern of heat dissipation that was
always found during the study period. At night, when flows
are close to zero, ∆Toff is 0.5 to 1.0 °C higher than dT. In the
daytime, when sap flow densities are greater than 0.5 l dm –2
h –1, the differences become negligible. The negative differences measured for high flows of the SW probe did not fit with
the general pattern and were probably linked to differences
Figure 4. Difference between maximum raw signals (∆TMr) and the
mean of maximum corrected signals (∆TMc ) in relation to maximum
natural temperature gradients (dTmax) for the three probe azimuths.
y = 0.7065x – 0.009, r 2 = 0.48, n = 192.
Figure 5. Difference between the noncontinuous signal after 15 min of
cooling (∆Toff) and the natural temperature gradients (dT ) in relation
to corrected sap flux density. Derived from data in Figure 1.
in natural temperature gradients or sap flow rates between
Days 1 and 3 of the sequence.
Our tests showed that, at night, 1 h was usually necessary to
reach absolute temperature equilibrium following both the
heating and cooling phases (data not shown). However, a noncontinuous signal with a 1 h on/1 h off cycle is not practical because of the low time resolution.
Even without calibration, the difference between ∆Ton and
∆Toff in the 45 min on/15 min off cycle, called alternate signal
(∆Ta), appeared to account for natural temperature gradients
and to correlate with sap flux density (Figure 1). Analysis of
the seasonal course of the maximum values recorded during
the night (∆TMa) with the same process described in Figure 3,
confirmed the higher stability of the alternate signal compared
with the continuous signal (Figure 6). Variations in ∆TMa were
Figure 6. Deviation from means of maximum alternate thermal dissipation probe (TDP) signals (∆TMa ) recorded at night for the three
probe azimuths (N, SE and SW). The alternate signal is derived from
the cyclic heating system. Means (± SD) for the N, SE and SW probes,
respectively, were: 8.8 ± 0.06, 8.2 ± 0.04 and 8.3 ± 0.06 °C.
TREE PHYSIOLOGY VOLUME 22, 2002
SAP FLOW MEASUREMENT WITH THERMAL DISSIPATION PROBES
small, between 0.08 and 0.12 °C. The alternate signal also
showed some similarities with the corrected continuous signal. Absolute means had the same ranking among probes; i.e.,
N > SW > SE (see captions of Figures 3 and 6). Otherwise, the
relationship between the deviations from means confirmed an
overall seasonal trend despite the 1-day lag between recordings (r 2 = 0.30, Figure 7); this trend was attributed to the dynamics of nighttime sap flow. The regression slope may also
be interpreted in terms of signal sensitivity: it was three times
smaller for the alternate signal than for the continuous signal.
Discussion
Importance of natural temperature gradients
Results obtained with continuously heated TDP probes
showed a direct influence of natural temperature gradients on
sap flow measurements with large potential errors. When considering the relative importance of such gradients, we need
to determine a threshold value beyond which gradients are no
longer negligible. The threshold can be estimated from
Granier’s (1985) empirical relationship (Equation 1) by considering the natural temperature gradient as additive to the gradient caused by heating. The threshold depends on actual sap
flow rate and maximum acceptable error. Taking a maximum
error of 10% and a zero flux signal of 10 °C, the threshold is
0.13 °C for a low sap flow rate of 0.5 l dm –2 h –1, and it reaches
0.35 °C for a high sap flow rate of 3.5 l dm –2 h –1. These estimates are valid for nighttime or daytime gradients. As a first
approximation, we can consider natural temperature gradients
in excess of 0.2 °C as non-negligible.
We found that natural temperature gradients were rarely
negligible. Although the N probe had some days with maximum nighttime gradients less than 0.2 °C, the daytime gradients had amplitudes of more than 0.3 °C. The importance and
effect of natural temperature gradients are specific to particu-
647
lar environmental conditions; they are amplified by the combination of a tree population of low and heterogeneous density,
high incident solar radiation and low sap flux density. Situations where natural temperature gradients are slightly greater
than 0.2 °C are probably common, making it essential to determine the natural temperature gradients by taking measurements with an unheated probe before starting measurements at
a new site.
Limits of gradient correction methods
The effects of natural temperature gradients on sap flow measurements have also been noted in temperate regions, and several methods of correction have been proposed (Èermák and
Kuèera 1981, Cabibel and Do 1991, Goulden and Field 1994,
Köstner et al. 1998). The simplest method of correction is to
record natural temperature gradients on the day preceding or
following the measurement day and then subtract these from
the temperature gradients measured during heating. This approach requires repeated sacrifices of measurement days and
assumes that gradients between consecutive days are equivalent, which is rarely the case because of possible changes in
pedo-climatic conditions and sap flows. To minimize this difficulty, Cabibel and Do (1991) used correlations between natural gradients for each probe and climatic data; however, this
process is complicated and unwieldy (Braun and Schmid
1999). Moreover, in our study, where local factors such as azimuth or sap flow (temperature and rate) appear to be important, the correlations were not precise enough for predictive
use.
Other correction methods use unheated blank probes that
are run and recorded simultaneously with heated probes.
Blank probes can be placed on the same trunk, necessarily at a
different position, or on individual trees serving as controls. In
the trunk, correction can be automatic, as in the design of
Goulden and Field (1994), who reused the compensation
method of Èermák and Kuèera (1981): one or two pairs of
thermocouples are integrated in the same electrical circuit of
the heated thermocouples. However, these methods require
that temperature gradients are homogeneous around the trunk
or between trees, a scenario which is not supported by our
study. The compensation method may take into account some
spatial variability by testing blank probes at different positions, but the problem of temporal variability remains. The
preceding methods are useful in certain situations, but by definition have limited application and should be frequently
tested.
Benefits of a noncontinuous TDP signal
Deviation of maximum corrected thermal dissipation probe (TDP)
signals (∆TMc) versus deviation of maximum alternate signals (∆TMa )
for the three probe azimuths. y = 2.678x + 0.001, r 2 = 0.30, n = 192.
In contrast to the continuous TDP system, the alternate signal
from the cyclic heating regime (45/15 cycle of heating and
cooling) appears to take natural temperature gradients into account. The precise mechanism is unknown and has not been
studied, but the long-term nighttime stability confirms that
there is little, if any, sensitivity to natural temperature gradients. The alternate signal is attractive because it may provide a
more general solution to natural temperature gradients that is
linked to the principle underlying the functioning of TDP
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
648
DO AND ROCHETEAU
probes. However, the alternate signal is different from the continuous TDP signal corrected by natural gradients because the
time schedule of heating and cooling does not allow probes to
reach a complete thermal steady state. Hence, a new calibration study is necessary. A limitation to the use of an alternate
signal is the low time resolution. It is not yet known to what extent it is possible to reduce the timing of the measurement
cycle. We need to study noncontinuous TDP measurements
under conditions where temperature gradients, flux density
and the time schedule of heating and cooling can be controlled.
Analysis of nighttime signals revealed another constraint of
the TDP system, namely that it cannot assess zero flow. Zero
flow must be assumed during certain nights. External information or manipulations should be used to test the assumption. In
our study, we used the natural defoliation of the tree. Zero flux
assumption and sensitivity to natural temperature gradients
are problems common to other sap flow methods such as the
heat balance methods of Èermák et al. (1973) and Sakuratani
(1981). Heat pulse methods (Marshall 1958, Burgess et al.
2000) are less affected by temperature gradients and can assess zero sap flow and flow direction, but they are more
complex than thermal dissipation methods.
Conclusions
When using the thermal dissipation probe with continuous
heating, it is important to consider the effects of natural temperature gradients. Measurement errors associated with
temperature gradients are enhanced at sites with a large temperature difference between day and night, low rates of sap
flow, open stands or high solar radiation. Alternative methods
of correction have been proposed, but these are unsuitable
where high spatial and temporal variability of temperature
gradients is present. A noncontinuous thermal dissipation system is an attractive solution because it retains the simplicity of
Granier’s (1985) original method and apparently eliminates
the sensitivity to natural temperature gradients.
Acknowledgments
Research was funded by the Savanna program of IRD (Institut
français de Recherche pour le Développement, ex ORSTOM) and by
the INCO-DC program No. CT98 0322 “Minimising Competition in
Dryland Agroforestry.” We thank the Ecology laboratory team of
Dakar IRD center for the technical support. We especially thank
A. Granier and J.P. Lhomme for comments on the manuscript, and
D. Deans for English revision.
References
Allen, S.J. and V.L. Grime. 1995. Measurements of transpiration from
savannah shrubs using sap flow gauges. Agric. For. Meteorol. 75:
23–41.
Andrade, J.A., F.C. Meinzer, G. Goldstein, N.M. Holbrook, J. Cavelier, P. Jackson and K. Silvera. 1998. Regulation of water flux
through trunks, branches, and leaves in trees of a lowland tropical
forest. Oecologia 115:463–471.
Braun, P. and J. Schmid. 1999. Sap flow measurements in grapevines
(Vitis vinifera L.). 2. Granier measurements. Plant Soil 215:47–55.
Brenner, A.J., P.G. Jarvis and R.J. Vandenbeldt. 1991. Transpiration
from a neem windbreak in the Sahel. In Soil Water Balance in the
Sudano-Sahelian Zone. Proc. Niamey Workshop, IAHS Publ. No.
199, pp 375–385.
Burgess, S.S.O., M.A. Adams and T.M. Bleby. 2000. Measurements
of sap flow in roo ts of woody plants: a commentary. Tree Physiol.
20:909–913.
Cabibel, B. and F. Do. 1991. Mesures thermiques des flux de sève
dans les troncs et les racines et fonctionnement hydrique des
arbres: I. Analyse théorique des erreurs sur la mesure des flux et
validation des mesures en présence de gradients thermiques extérieurs. Agronomie 11:669–678.
Èermák, J. and J. Kuèera. 1981. The compensation of natural temperature gradient at the measuring point during the sap flow rate determination in trees. Biol. Plant. 23:469–471.
Èermák J., M. Deml and J.M. Penka. 1973. A new method of sap flow
rate determination in trees. Biol. Plant. 15:171–178.
Goulden, M.L. and C.B. Field. 1994. Three methods for monitoring
the gas exchange of individual tree canopies: ventilated-chamber,
sap-flow and Penman-Monteith measurements on evergreen oaks.
Funct. Ecol. 8:125–135.
Granier, A. 1985. Une nouvelle méthode pour la mesure des flux de
sève dans le tronc des arbres. Ann. Sci. For. 42:193–200.
Granier, A. 1987. Evaluation of transpiration in a Douglas-fir stand by
means of sap flow measurements. Tree Physiol. 3:309–319.
Guttiérrez, M.V., R.A. Harrington, F.C. Meinzer and J.H. Fownes.
1994. The effect of environmentally induced stem temperature gradients on transpiration estimates from the heat balance method in
two tropical woody species. Tree Physiol. 14:179–190.
Köstner, B., A. Granier and J. Èermák. 1998. Sapflow measurements
in forest stands: methods and uncertainties. Ann. Sci. For. 55:
13–27.
Le Houérou, H.N. 1989. The grazing land ecosystems of the African
Sahel. Ecological Studies. Vol. 75. Springer-Verlag, Heidelberg,
282 p.
Marshall, D.C. 1958. Measurement of sap flow in conifers by heat
transport. Plant Physiol. 33:385–396.
Roupsard, O., A. Ferhi, A. Granier, F. Pallo, D. Depommier, B. Mallet, H.I. Joly and E. Dreyer. 1998. Fonctionnement hydrique et profondeur de prélèvement de l‘eau de Faidherbia albida dans un parc
agroforestier soudanien. In L’Acacia au Sénégal. Eds. C. Nef
Campa, S. Hamon and C. Grignon. Collection Colloques and
Séminaires, ORSTOM Editions, Paris, pp 81–103.
Sakuratani, T. 1981. A heat balance method for measuring water flux
in the stem of intact plants. J. Agric. Meteorol. 37:9–17.
Smith, D.M. and S.J. Allen. 1996. Measurement of sap flow in plant
stems. J. Exp. Bot. 47:1833–1844.
Wullschleger, S.T., F.C. Meinzer and R.A. Vertessy. 1998. A review
of whole-plant water use studies in trees. Tree Physiol. 18:
499–512.
TREE PHYSIOLOGY VOLUME 22, 2002