Journal of Experimental Botany, Vol. 47, No. 305, pp. 1957-1962, December 1996
Journal of
Experimental
Botany
Simultaneous measurements of transpiration, soil
evaporation and total evaporation in a maize field in
northern Germany
M. Herbst 13 , L. Kappen 12 - 4 , F. Thamm 1 and R. Vanselow1
1
Ecosystem Research Center, University of Kiel, Schauenburgerstr. 112, D-24118Kiel, Germany
2
Botanical Institute, University of Kiel, Olshausenstr. 40, D-24098 Kiel, Germany
Received 21 September 1995; Accepted 3 June 1996
Abstract
Transpiration and evaporation in a maize field were
measured by porometer, lysimeter and Bowen ratio
techniques. All methods provide corresponding and
complementary data for daily evaporation sums. It is
proposed to use these techniques simultaneously in
order to estimate reliably the partitioning of evaporation between plants and soil.
Key words: Transpiration, soil evaporation, total evaporation, partitioning, maize.
Introduction
Plant water utilization in agricultural land and its effects
on the yield of agricultural crops can be explained only
from the partitioning of total evaporation (Et) into transpiration (T), soil evaporation (£,) and interception evaporation (£•,). Therefore, modelling approaches based on
the concept of physical and physiological surface resistances for water vapour transport are more and more
replacing simple approaches using potential evaporation
rates combined with empirical factors if actual evaporation from agroecosystems has to be predicted. In particular, the use of dual source models, recognizing the
contributions of plants and soil to total evaporation,
increases, because these models seem to be applicable to
a wide range of conditions (Wallace, 1995). Their parametrization requires an independent investigation of soil and
plant evaporation. Until now, transpiration has mostly
been measured by plant physiologists who have used
porometers and provided data on the leaf scale. Soil
3
4
evaporation beneath plant canopies has mostly been
investigated by hydrologists using lysimeters in small
plots of soil. Total evaporation from soil and plants on
the stand scale has often been determined by meteorologists who apply atmospheric methods like the Bowen
ratio-energy balance approach or the eddy covariance
technique. However, results of these independent
approaches have rarely been brought together. Et measured by atmospheric methods can equal T if the plant
canopy under investigation is closed and equals E, over
bare soil. In sparse crops, such as on fields during the
early growth stages of plants, stand scale methods provide
a mixture of plant and soil evaporation rates. To get an
information about the partitioning of evaporation
between plants and soil, measurements have to consist of
a combination of at least two of the methods mentioned
above.
Some experiments have been reported where either Et
and T or £, and E, were measured directly, and where E,
or T, respectively, were calculated as the difference
between the other two quantities, but not validated by
direct measurements (Sakuratani, 1987; Ham et al., 1991;
Ashktorab et al., 1994). To our knowledge, the only
study where both transpiration and evaporation as well
as the sum of the two were measured directly is reported
by Wallace et al. (1993), who used eddy covariance,
porometer and lysimeter techniques simultaneously in a
sparse dryland crop.
Since the significance of evaporation partitioning has
been widely emphasized (Ashktorab et al., 1989; Leuning
et al., 1994), an experimental approach has been applied
for the first time in an agricultural crop in a humid-
To whom correspondence should be addressed. Fax: +49 431 880 4083. E-mail: [email protected]
Dedicated to Professor Dr Reinhard Bomkamm on the occasion of his 65th birthday.
6 Oxford University Press 1996
1958
Herbst et al.
temperate region, with the simultaneous usage of three
techniques for the investigation of transpiration and
evaporation in a maize field. A Bowen ratio system and
a weighing lysimeter operated continuously and provided
data for El and E%, respectively. The difference between
Et and Et was T if the canopy was dry or T+E, if the
canopy was wet, respectively. Additionally, a porometer
was used to validate T for selected days. With this paper
(i)
a simple experimental design to measure plant, soil
and total evaporation directly and simultaneously is
proposed,
(ii) some examples of data obtained in the 1995 growing
season in a maize (Zea mays) field in northern
Germany are given, and
(iii) attention is drawn to some general problems involved
in comparing data collected on different scales.
Materials and methods
Study site
The investigations were carried out in the study area of the
interdisciplinary project 'Ecosystem Research in the BornhSved
Lakes region' 30 km south of Kiel (northern Germany), at
54°06' N and 10° 15' E, in a region with a marine, humidtemperate climate. The investigated maize field covers 2.5 ha
and is located on a typical cambisol, developed on loamy to
silty moraine sand topping fluvioglacial sand (Schleuss, 1992).
In early May 1995 maize (cvs Diamant and Helix) was sown at
a density of 8.75 plants m~2, in rows spaced 0.75 m. In late
July the plants had reached maximum leaf area index (LAI) of
2.5 and started flowering. During August an extended period
of drought caused a decrease of LAI.
Measurements of transpiration
Maize transpiration was calculated from leaf stomatal conductances (g,) measured with a steady-state porometer LI-1600
(Li-Cor, Lincoln, USA) on leaves of three canopy layers,
representing the upper 25%, middle 50% and lower 25% of the
canopy (Rochette et al., 1991). Because maize has amphistomatous leaves, g, values obtained for adaxial and abaxial leaf
surfaces had to be added together. These data were scaled up
to a 'canopy conductance' gc by use of LA I data of representative
leaves. LAI was calculated by means of an empirical relationship
between length, width and surface area of the maize leaves (R
Vanselow, unpublished data). T was finally calculated from the
canopy conductance and stand microclimate by use of the
Penman-Monteith equation:
XE=
sxA+pxcpx Dxg,
(1)
where AE is latent heat flux, s is the slope of the curve relating
saturated vapour pressure to temperature, A is available energy
(i.e. the difference between net radiation and soil heat flux), p
is density of dry air, cp is specific heat of air, D is saturation
deficit of the air above the canopy, g% is the aerodynamic
conductance between the canopy level and the reference height,
and y is the psychrometric 'constant'.
Net radiation was measured with a tube net radiometer
(TRL, Delta-T Devices Ltd., Cambridge, UK) placed 4 m
above the ground, and soil heat flux with CN3 heat flux plates
(Thies GmbH, G6ttingen, Germany). Air temperature and
humidity above the plant canopy were determined by use of
HMP35A humidity sensors (Vaisala, Helsinki, Finland).
To calculate g,, a formula given by Thorn (1975) was used
(eqn. 2),
k2xu(z)
In ((z-dj/zj
2
(2)
where k is von Karman's constant, and z is the reference height
of the energy balance measurements. Wind speed above the
canopy u(z) was measured with a cup anemometer of type
'LISA' (Siggelkow GmbH, Hamburg, Germany). Roughness
length z0 as well as zero plane displacement height d were
estimated from stand height and leaf area distribution as
described by Shaw and Pereira (1982).
Measurements of soil evaporation
Soil evaporation was measured with a continuously weighing
mini-lysimeter that was placed in the middle between two rows
of the maize plants shortly after their germination. The lysimeter
was 20 cm deep and had a diameter of 26 cm, corresponding to
a surface area of 530 cm2. The soil water potential in the
lysimeter vessel was achieved by artificial suction to be equal
to that in the ambient soil. The walls of the lysimeter were
made of PVC, the suction plates in the bottom contained a
membrane consisting of nylon and polyester (Ultipor N66, Pall
Filtrationstechnik, Dreieich, Germany). The load cell (type
DBBP, Althen, Kelkheim, Germany) is temperaturecompensated and has a capacity of 20 kg. In connection with
the lysimeter surface area it is sensitive to approximately 0.1 mm
water. Percolation was collected in a U-shaped tube, continuously registered by a pressure sensor (type PDCR 800, Druck,
Bad Nauheim, Germany) and subtracted from the weight loss
value of the lysimeter to get £,. Further technical details are
given by Thamm (1993). The data were averaged bi-hourly
because hourly evaporation rates were sometimes smaller than
the resolution limit of the weighing lysimeter.
Measurements of total evaporation
Total evaporation of the stand was determined with the Bowen
ratio—energy balance method, which calculates the latent heat
flux from measurements of the canopy energy balance and of
temperature and humidity gradients above the canopy (Bowen,
1926). The equations of the energy balance (eqn. 3) and of the
ratio of sensible and latent heat fluxes (Bowen ratio, eqn. 4)
are solved simultaneously, neglecting the metabolic heat flux
and assuming equal eddy diffusivities for sensible heat and
water vapour. The energy balance for a canopy between a
reference level above and the soil surface is:
N
+ G+H+XE+M=0
(3)
where RN is net radiation, G is soil heat flux, H and XE are
turbulent fluxes of sensible and latent heat, respectively, A is
latent heat of vaporization of water, E is the evaporation or
transpiration rate, and M is metabolic heat flux. Allfluxesare
defined as positive downward.
The Bowen ratio (/3) is:
H
KH
AT
—
Ae
(4)
where AT and Ae are vertical temperature and humidity
differences over the same height interval above the canopy,
Simultaneous measurement of transpiration and evaporation
respectively, KH and KE are eddy diffusivities for heat and water
vapour, respectively.
Neglecting M and assuming A"H = KE,
(5)
A£=
1+jB
Net radiation, soil heat flux, and temperature and vapour
pressure differences between 0.1 and 1.1m above the crop were
measured as described above. Since humidity differences were
often smaller than the calibration accuracy of the sensors, a
system was constructed following the method of Cellier and
Olioso (1993) containing only one capacitive humidity sensor
(HMP35A, Vaisala, Helsinki, see Fig. 1). Air samples from
both measurement heights were pumped separately to the sensor
by two pumps alternately operating every 5 min. Instead of
valves, a switch was installed (Fig. 1) to prevent air passing
through the non-operating pump. Measurements were taken
every 30 s and averaged hourly. During the time lag (5 min)
between the measurements at the two levels there may sometimes
occur changes of ambient humidity ranging in the same order
of magnitude as the gradient between the two measurement
levels itself. Therefore, the observed gradients were corrected
by the average humidity courses during the time lag. The air
tubes were heated periodically to avoid condensation of water
in the tubes. As a result, vapour pressure (not relative humidity)
at the two measurement heights is recorded with this Bowen
ratio system.
Theoretically, the Bowen ratio method is restricted to flat,
homogeneous terrain of infinite extension, but in practice it has
proven to be useful in many non-ideal situations as well
(Ashktorab et al., 1989; Nie et al., 1992). The most crucial
-<
Pump
Pump I
Air Intakes
at two heights
Fig. 1. The system used to measure the Bowen ratio. It consists of a
PVC box with two pumps, a humidity sensor (Vaisala HMP35A) and
a FT 100 temperature sensor. See text for further explanations.
1959
parameter for the application of this method is the upwind
extension of the canopy to be measured (fetch) in relation to
the height of the sensors above the evaporating surfaces.
Because the micrometeorological measurements were carried
out near the southern and western edges of the field, data from
periods with these wind directions had to be excluded from any
further analysis. If wind was between north and east, fetch-toheight-ratios ranged between 50:1 and 180:1, which can be
considered as being sufficient for measurements over smooth
surfaces like agricultural crops (Gash, 1986; Heilman et al.,
1989; Nie et al., 1992).
Results
Performance of the systems
The Bowen ratio and lysimeter systems worked continuously throughout the growing season (mid-May to early
October) and required half a day every week for maintainance and/or exchange of sensors and pumps, and for
control of the lysimeter suction. Data observed by the
Bowen ratio method were reliable as long as the fetch
was sufficient and there was no rainfall. Field operations
with the porometer and L/l/-estimates were very timeconsuming and could be carried out for selected rainand fogless days only. The performance of the LI-1600
porometer in the field was reliable, although care had
always to be taken to prevent warming of the cuvette
above ambient temperature, which could affect stomatal
behaviour of the leaves.
Diurnal courses of E, T, and E,
On 25 May, shortly after germination of maize and 6 d
after the last rain, transpiration was still effectively zero.
The soil had been dried out in the uppermost 2-3 mm.
Although the daily courses of soil evaporation exhibit
slight discrepancies between lysimeter and Bowen ratio
data (Fig. 2), the calculated evaporation totals on this
cloudy day were identical (1.39 mm d" 1 ).
On 26 June, a cloudless day 4 d after the last rain, the
maize plants had grown up to 50 cm height. Cumulative
LAI was approximately 0.6. Leaf stomatal conductance
as influenced by both light and humidity conditions did
not vary much during the daylight period. Soil was dry
in the uppermost 2 cm and lost only a small amount of
water (0.31 mm d~') by evaporation. The total evaporation per day obtained by the Bowen ratio method
(2.50 mm d" 1 ) approximately equalled the sum of T and
E, (2.57 mm d" 1 ). The deviation between hourly data is
sometimes considerable, especially in the early morning
hours and around noon, but exhibits no systematic trend.
20 July was the second rainless day following a period
of four rainy days. Soil surface was still wet at the
beginning of the day. Maize plants had almost reached
their maximum LAI and were rapidly growing (up to
8 cm d" 1 ). Stomatal conductances on this day were the
highest observed during the 1995 vegetation period.
1960 Henbsfetal.
[W/m
2,
600
400
200
18 August
0
H
1500
o
E
_E
'000
LAI = 0.0
1
h-
LAI = 2.2
500 -
I "T^H—
n
Q0
JC
\
E
E
6
9
12
15
18
6
9
12
15
18
-
i
I Soil Evaporation (Lysimeter),
Transpiration (Porometer),
-
Evapotranspiration (Bowen Ratio)
Fig. 2. Diurnal courses of net radiation (Z?N), vapour pressure difference (VPD), leaf stomatal conductance (g,), and of plant, soil and total
evaporation (E) as measured by porometer, lysimeter and Bowen ratio techniques, respectively. Daily totals of evaporation are shown in the
lower panels.
Although the sky was cloudy during most of the day, net
radiation input was, on average, only 30% lower than
during clear days. VPD did not exceed 1.5 kPa, and
windspeed was very low. Since wind directions changed
several times during the day, hours with critical wind
directions were excluded from the Bowen ratio calculations. Including linearly interpolated values for the missing hours, Et was estimated as 6.24 mm d" 1 which is
close to the value of 6.17 mm d" 1 as the sum of T
(4.79 mm d" 1 ) and E, (1.38 mm d" 1 ).
On 18 August, subsequent to a period of four hot and
rainless weeks, soil evaporation was effectively zero. The
missing water supply to the plant roots in combination
with an extremely high vapour pressure difference of the
air around 3 kPa caused nearly complete stomatal closure
from 10:00 a.m. until evening. Transpiration sums for
this day obtained by porometry (1.69 mm d" 1 ) and the
Bowen ratio method (1.46 mm d ~') differed by approximately 15%. Deviations between both methods were
largest during the morning hours when temperature and
humidity gradients were small and stomatal reactions
took place rapidly, as a response to rapidly changing
ambient conditions.
Discussion
Error assessment
Detailed error analyses for the Bowen ratio technique
have been presented by Sinclair et al. (1975) and Angus
and Watts (1984). Because inaccuracies in determining
the temperature and humidity differences were minimized
in this study by using one sensor only, the probable errors
of this method can be traced to the measurement of net
radiation and soil heat flux. The accuracies of both the
tube net radiometers and the soil heat flux plates are 5%
according to the manufacturers. The error for G measured
in the field can be somewhat larger due to spatial variability of soil characteristics and of radiation input beneath
the plant canopy, say 30%. However, because during the
day G was never more than 30% of 7?N in bare soil and
15% of Rs under the closed canopy, the resulting error
Simultaneous measurement of transpiration and evaporation
of A (difference of RN and G) and consequently of XE
can be estimated to be only 10% for normal, midday
conditions (see also Sinclair et al., 1975).
With respect to lysimetry, inaccuracies may arise from
the combination of load cell resolution limit and lysimeter
size (see 'Materials and methods'), as well as from problems involved with representativeness of the lysimeter
measurements: As soil evaporation depends on moisture
content of the upper soil layers, it is affected by the
amount of precipitation reaching the soil surface, which
varies with distance from the rows (Leuning et al., 1994).
This may be of importance in the present study because
the diameter of the lysimeter was much smaller than the
row-to-row distance. Furthermore, the lysimeter vessel
contained bare soil only, so that the amount of soil water
that was extracted by the plant roots had to be simulated
by artificial suction, which also may have been inaccurate
at some times. Thus, it can not be excluded that E,
obtained from lysimeter data may have an uncertainty of
up to 50% for a vegetated, wet soil surface. In bare soil,
the accuracy of this technique should be much better.
Several sources of errors in the determination of stand
transpiration from measurements of leaf stomatal conductances have to be considered. First, the accuracy of
the LI-1600 porometer with respect to stomatal conductance can be assumed to be 10%, according to the manufacturer. Second, the systematic spatial variability of
stomatal reactions over a leaf, a plant and a stand of
plants can be considerable (Leverenz et al., 1982; Rochette
et al., 1991). Even if data are sampled in different layers
of the canopy as in the present study, an additional
uncertainty in the determination of gc of say 20% has to
be taken into account. Third, a non-destructive estimation
of LAI may cause an additional uncertainty in the order
of 10% (Eschenbach and Kappen, 1996) for gc. Finally,
some parameters in the Penman-Monteith equation (A,
D, gB) may be in error by a few per cent as well, although
their influence on XE in this non-linear equation is not as
strong as that of gc. Calculating A£ simply as the product
of gc and D would result in additional errors because of
neglecting leaf-to-air temperature differences and aerodynamic resistances to water vapour transfer. Thus, if all
sources of error mentioned would act together in the
same direction, stand transpiration calculated by this
method could be in error by perhaps 40 or 50%.
Fortunately, this will not usually be the case from a
statistical point of view, therefore a typical uncertainty
of about 20% may reasonably be assumed for stand
transpiration values derived from porometry.
Most of the named error sources are not linked to each
other and can therefore operate independently in both
directions. Thus it can be concluded that the agreement
of the methods will improve with increasing sampling
density in time and space.
1961
Interpretation of experimental data
The daily sums of Et measured by the Bowen ratio
technique and of Tplus E, measured with porometer and
lysimeter differed by less than 15% in the present study.
A similar agreement between these techniques in general
was observed by Wallace et al. (1993), who estimated T,
£, and Et from a millet crop grown in a semi-arid
environment by use of porometer, lysimeter and eddy
covariance techniques. An overestimation of T determined by porometers at high LAI by up to 60% for daily
totals found by these authors was not observed in our
study. However, some hourly Et values deviated significantly from T+E,, which could have resulted from several
reasons not yet discussed in the previous section about
general uncertainties of the techniques used.
Whilst porometer data are collected on a momentary
base, Bowen ratio and lysimeter techniques provide
hourly (or bi-hourly, respectively,) averages from readings
every minute. This can cause discrepancies if environmental conditions and, subsequently, stomatal reactions
change rapidly as is the case shortly after sunrise, for
instance. The morning hours anyway seem to be the most
problematic times for employing the proposed techniques,
due to the possibility of dew formation on leaves and
instruments. Additionally, the Bowen ratio technique does
not function when temperature and humidity gradients
reverse in the morning. Small errors in the determination
of gradients (even if minimized by the use of only one
humidity sensor) as well as the continuing adjustment of
the gradients may then lead to large errors in evaporation
rates as was shown by Angus and Watts (1984). Finally,
porometer measurements are point measurements not
only in time but also in space. The heterogeneity of plant
canopies strongly affects the scaling procedures from leaf
to canopy (Leverenz et al., 1982). In the experiment,
porometer measurements were carried out only in a small
plot of the maize field for practical reasons. Any comparison of upscaled porometer values with stand scale observations is highly sensitive to the distribution of LAI and
plant physiological and soil characteristics over the whole
field as it is 'seen' by the Bowen ratio sensors. It can not
be excluded, for example, that the relatively low values
of stand evaporation during midday of 26 June and in
the morning of 18 August corresponded to a real difference in stomatal behaviour between the maize plants
measured with the porometer and those upwind of the
instrument mast.
Conclusions
Does the observation of complementary data gained by
three different techniques allow the conclusion that the
employment of any two of the methods described is
adequate to estimate evaporation partitioning? Cases may
1962 Herbst etal
come up where a particular technique can not be applied,
e.g. the Bowen ratio in very small experimental plots, or
where porometry and determination of LAI are too
complicated because of a strong heterogeneity within the
canopy. Then the application either of porometry plus
lysimetry or of Bowen ratio plus lysimetry may give
plausible results explaining evaporation partitioning, at
least over time periods > 1 d, but remaining uncertainties
in the order of 15% or 20% have to be accepted even on
this time scale. In fact, the results of this study demonstrate that the direct, simultaneous measurement of £„ T
and E, on different scales can make statements about
evaporation components in agricultural land more certain, because data collected on the leaf scale can be tested
against stand scale data or vice versa. Employing the
three methods simultaneously, Bowen ratio, lysimeter and
porometer, therefore facilitates the detection of doubtful
or erroneous data and allows a most reliable estimate of
the partitioning of evaporative water loss between plants
and soil. This may have significance for both humidtemperate and arid regions to predict, for instance, to
what extent crop yield and water balance will be affected
by climate change.
Acknowledgements
This work was earned out in the framework of the interdisciplinary project 'Ecosystem research in the BornhSved lakes region'
which was funded by the Ministry of Research and Technology
of the Federal Republic of Germany. We thank Helge
Petrikovsky and Ronald Stenzel for their work on the design
of the Bowen ratio system and Thorsten BSker and Holger
Michling for their support during the field work.
References
Angus DE, Watts PJ. 1984. Evapotranspiration—how good is
the Bowen ratio method? Agricultural Water Management
8, 133-50.
Ashktorab H, Pruitt WO, Paw U KT. 1994. Partitioning of
evaporation using lysimeter and micro-Bowen ratio systems.
Journal of Irrigation and Drainage Engineering 120, 450-64.
Ashktorab H, Pruitt WO, Paw U KT, George WV. 1989. Energy
balance determinations close to the soil surface using a microBowen ratio system. Agricultural and Forest Meteorology
46, 259-74.
Bowen IS. 1926. The ratio of heat losses by conduction and by
evaporation from any water surface. Physical Review 27,
779-87.
Cellier P, Olloso A. 1993. A simple system for automated long-
term Bowen ratio measurement. Agricultural and Forest
Meteorology 66, 81-92.
Eschenbach C, Kappen L. 1996. Leaf area index determination
in an alder forest—a comparison of three methods. Journal
of Experimental Botany 41, 1457-62.
Gash JHC. 1986. Observations of turbulence downwind of a
forest-heath interface. Boundary-Layer Meteorology 36,
227-37.
Ham JM, Heilman JL, Lascano RJ. 1991. Soil and canopy
energy balances of a row crop at partial cover. Agronomy
Journal 83, 744-53.
Heilman JL, Brittin CL, Neale CMU. 1989. Fetch requirements
for Bowen ratio measurements of latent and sensible heat
fluxes. Agricultural and Forest Meteorology 44, 261-73.
Leuning R, Condon AG, Dunin FX, Zegelin S, Denmead OT.
1994. Rainfall interception and evaporation from soil below
a wheat canopy. Agricultural and Forest Meteorology 67,
221-38.
Leverenz J, Deans JD, Ford ED, Jarvis PG, Milne R, Whitebead
D. 1982. Systematic spatial variation of stomatal conductance
in a Sitka spruce plantation. Journal of Applied Ecology
19,835-51.
Nie D, Flitcroft ID, Kanemasu ET. 1992. Performance of Bowen
ratio systems on a slope. Agricultural and Forest Meteorology
59, 165-81.
Rochette P, Pattey E, Desjardins RL, Dwyer LM, Stewart DW,
Dube PA. 1991. Estimation of maize (Zea mays L.) canopy
conductance by scaling-up leaf stomatal conductance.
Agricultural and Forest Meteorology 54, 241-62.
Sakuratani T. 1987. Studies of evaporation from crops. II.
Separate estimation of transpiration and evaporation from a
soybean field without water shortage. Journal of Agricultural
Meteorology 42, 309-17.
Schleuss U. 1992. Boden und Bodenschaften einer norddeutschen Moranenlandschaft-Okologische Eigenschaften,
Vergesellschaftung und Funktionen der BOden im Bereich
der BornhSveder Seenkette. EcoSys (Kiel) Supplement 2,
1-185.
Shaw RH, Pereira AR. 1982. Aerodynamic roughness of a plant
canopy: a numerical experiment. Agricultural Meteorology
26, 51-65.
Sinclair TR, Allen LH, Lemon ER. 1975. An analysis of errors
in the calculation of energy flux densities above vegetation
by a Bowen ratio profile method.
Boundary-Layer
Meteorology 8, 129-39.
Thamm F.
1993. Kontinuierliche
Registrierung
der
Sickerwassermenge aus der Humusauflage - Vorschlag zur
technischen Realisierung. Mitteilungen der Deutschen
Bodenkundlichen Gesellschaft 72, 265-8.
Thorn AS. 1975. Momentum, mass and heat exchange of plant
communities. In: Monteith JL, ed. Vegetation and the
atmosphere, Vol.1. London: Academic Press, 57-109.
Wallace JS. 1995. Calculating evaporation: resistance to factors.
Agricultural and Forest Meteorology 73, 353-66.
Wallace JS, Lloyd CR, Sivakiimar MVK. 1993. Measurements
of soil, plant and total evaporation from millet in Nigeria.
Agricultural and Forest Meteorology 63, 149-69.