Professional Documents
Culture Documents
Lysimeter Mikro - Boast 1982
Lysimeter Mikro - Boast 1982
Laboratory Evaluation1
C. W. BOAST AND T. M. ROBERTSONZ
water content (gm H20/gm dry soil) run 6 "dry" initial condition
0.00 o. 10 0.20 0.3O
• e v a p o r a t i v i t y = 2.1 mm/day
——i——i——i——i——i——i——i——i——i——i——i——i——i——i——|—
• 4» T • m -O V CS<3
runs
_^_^ •• TA • CDO£7 M
o 1
E A 2 I 2.0--
E 5O- V 3. 4 • • TA <EDdt»
• 5. 6 o
.c A 7. 3 XI
a. T 9, 10 •4 m O VE> W«
a>
• 11 . 12
•01 OO-
• 13. 14 11.5
< 15, 16
I 0 17. 18 «^K£SD7 •
D 19, 20
15O- O 21. 22
<=1.0
Fig. 2—Initial soil water content for the 12 micro-lysimeter samplings. o
Open and filled symbols distinguish samples which are classified • water container!
as having wet and dry initial conditions, respectively. A 146-mm m l c r o - l f » i m « t e r i
V 106-mm m l c r o - l y s l m e t e r i 1 i m e i of
of samples were wet with values between 0.23 and weighing, )|m .
0.5-- O 70-mm m l c r o - l y i i m « t « r i
period)
0.29 (open symbols). At the time of sampling the soil D 44-mm m l c r o - l y t t m e t e r i
surface appeared dry and wet, respectively, for the
two groups of samples.
As shown in Fig. 2, the difference in surface soil 0.0'
water content extends to 146 mm, the depth of the 1 2
longest micro-lysimeter sample, except for one set of t i m e ( d a y s ) —J>
samples (the filled circles).The samples represented Fig. 3—Rate of evaporation for a low evaporativity, dry initial con-
by filled circles were taken when the soil was frozen dition run.
at about the 100-mm depth and provide the oppor-
tunity to examine the behavior of soil in which there testable conclusions. However, for reasons of clarity,
is a sharp increase in water content at about the 50- the run portrayed in Fig. 3 is used as an example in
mm depth. One would expect the evaporation from the following exposition of the method of data analysis.
short micro-lysimeters to deviate from reality very It is assumed in this study that an "infinitely long"
quickly if there is a considerable amount of water micro-lysimeter would behave the same as undis-
available for upward flow from just below the bottom turbed soil and that micro-lysimeters of finite length
of the short micro-lysimeters. behave, for a time, as if they were infinitely long. The
most obvious way to know whether micro-lysimeters
Interpretation of Individual Runs of one length have ceased to behave like infinitely
The evaporation data for a run of low initial surface long ones is to compare them with much longer micro-
soil wetness and low evaporativity (2.1 mm/d) are lysimeters. However, even a very long micro-lysim-
illustrated in Fig. 3. The rate of evaporation from the eter will eventually deviate from infinitely long be-
micro-lysimeters for time period 1 (between the first havior. The problem is to know when this occurs.
two weighings) is essentially the same for all four Hence in this experiment, if at any time the next to
lengths. During the second time period the average the longest micro-lysimeters are still evaporating at
rate of evaporation from the 44-mm micro-lysimeters the same rate as the longest ones, this is taken to
falls below that of the longer ones. This occurs for indicate that both of these sets of micro-lysimeters are
the 70-mm micro-lysimeters during the third time pe- still behaving as infinitely long ones. For example, the
riod, and for the 106-mm ones during the sixth time 106- and 146-mm micro-lysimeters of Fig. 3 are re-
period. garded as standards for the 70- and 44-mm micro-
Figure 3 is presented to illustrate the experimental lysimeters for the first five time periods (2.2 d). After
procedure, not as typical data; of the 22 runs made, this time, when the 106-mm micro-lysimeters begin
it is atypical in that it best approximates "model" to deviate from the 146-mm ones, there is no way to
behavior. However, the variability exhibited is typi- know whether the 146-mm micro-lysimeters are still
cal. Note that if both the 146-mm micro-lysimeters behaving as infinitely long ones, although for a while
(upward-pointing triangles) had behaved like the one they probably are.
which evaporated faster, then the data would imply Establishing a criterion for when the 106- and 146-
that the evaporation rate of the three shorter-length mm micro-lysimeters are behaving as infinitely long
micro-lysimeters was never as large as that of the 146- ones is only part of the task, however. The goal of
mm-long ones. Conversely if both 146-mm micro-ly- the experiment is not to determine when short micro-
simeters had behaved like the one which evaporated lysimeters begin to deviate from infinitely long be-
slower, then the data would imply that the shorter havior, but to quantify the error in estimating field
micro-lysimeters were evaporating faster than the 146- evaporation using relatively short micro-lysimeters.
mm ones for the first two time periods. For example, in Fig. 3 at the end of the fifth time
In light of this variability no conclusions are drawn period, cumulative evaporation from the 70-mm mi-
directly from the data of any single run; data from cro-lysimeters is 0.45 mm less than the average for
more than one run are combined to draw statistically the 106- and 146-mm micro-lysimeters. This suggests
692 SOIL SCI. SOC. AM. J . , VOL. 46, 1982
E 8
£ 2--
o
T3
c
o
-O 1- -
— 6- -
o
a. -r-
o 20 44 70
micro-lys imeter
4- - 20-mm l e n g t h (mm)
O
Fig. 5—Time td at which short micro-lysimeters have evaporated 0.5
mm less than 106- and 146-mm micro-lysimeters for all of the low
^^ A H6-mm m i c r o - l y s l m e l e r s evaporativity wet initial condition runs. The three filled circles
2-- V 106-mm m i c r o - l y s i m e t e r s represent the td values determined in Fig. 4.
— - - t h e aboYe minus 0.5 mm
——— O 70-mm micro-lysimelers
——— D 44-mm m i c r o - l y s i m e t e r s a set amount (say 0.5 mm) from the longer ones.
r d = 0.8 However, this question can be discussed more clearly
days ——— O 20-mm m i c r o - l y s i m e t e r s
-f- after the variability of some "0.5-mm-deviation" data
2 6 is illustrated.
time ( d a y s ] In Fig. 5, the 0.5-mm-deviation times, td, deter-
Fig. 4—Cumulative evaporation for a low evaporativity, wet initial mined in Fig. 4 are plotted, together with similarly
condition run, including the definition of the 0.5-mm-deviation time determined times from the four other low-demand,
wet initial condition, runs. It was clear after three of
these five runs that the 20-mm micro-lysimeters de-
that a deviation of 0.45 mm occurs in the 70-mm micro- viated from the longer ones so rapidly that they would
lysimeters before uncertainty begins about the infi- be of little use in estimating field evaporation. Hence
nitely long status of the two longer sets of micro- 20-mm micro-lysimeters were not included in later
lysimeters. Unfortunately the data from most of the runs.
other runs are too variable for such a statement to be One cause of the variability in Fig. 5 is seen in the
made about them run by run. In the next section, scatter of the data in Fig. 4. For example, for the first
where all 22 runs are considered together, deviations .1.7 d after sampling, the curve representing the 70-
of up to 0.5 mm by the 70-mm micro-lysimeters are mm micro-lysimeters is above the thick solid line in
shown to occur before the 106-mm micro-lysimeters, Fig. 4. As a result, the determination of 3.7 d as the
taken as a whole, deviate from the 146-mm ones. cross-over time may be unrealistically high, and, in
Because of this, a deviation 0.5 mm is used (somewhat fact, it is the largest (filled circle) of the five values
aribtrarily) as a benchmark for evaluating micro-ly- shown for 70-mm micro-lysimeters in Fig. 5.
simeter behavior in the following analysis. In the previous section it was shown that the within-
Cumulative evaporation from micro-lysimeters of run variability makes it unproductive to ask, for each
the five lengths (146, 106, 70, 44, and 20 mm) over individual run, the question: do the 106- and 146-mm
a seven-day period is shown in Fig. 4 for a run of low micro-lysimeters evaporate at the same rate as each
evaporativity and wet initial condition. A thick solid other long enough to establish a deviation of 0.5 mm
line indicates the average cumulative evaporation for in the 70-mm micro-lysimeters? On the other hand,
the two longest sets of micro-lysimeters (triangles). the variability among runs makes it difficult to ask this
A dashed line is drawn 0.5 mm below the thick solid same question for all the runs taken as a whole. Con-
line. The point at which a cumulative evaporation sider, for example, the five data points for 70-mm
curve for one of the three short sets of micro-lysim- micro-lysimeters in Fig. 5. For the run represented
eters crosses the dashed line indicates the time, de- by the bottom data point the time in question is 1.9
noted td, at which the cumulative evaporation of the d, whereas for the top data point, it is 3.7 d. This
short micro-lysimeters deviates by 0.5 mm from that range pertains only to the runs with wet initial con-
of the 146- and 106-mm micro-lysimeters. In Fig. 4 ditions and relatively low evaporativity. If all 22 runs
for the 20-mm micro-lysimeters, td is 0.8 d. The 0.5- are considered, the range is wider (0.8 to 7.2 d). Thus,
mm-deviation times for the 44- and 70-mm micro-ly- although for any one run there is a time at which the
simeters are td = 1.8 and td = 3.7 d, respectively. question should be asked, for all the runs taken to-
gether there is no such single time.
Interpretation of Multiple Runs An approach to this variability dilemma is to con-
Logically the first question to address is whether sider differences in cumulative evaporation,mmnot as a
the 146- and 106-mm micro-lysimeters do indeed evap- function of time t, but as a function of tlt™~ , where
orate at the same rate as each other long enough to ^7o-mm denotes the value of td for the 70-mm micro-
establish that the shorter micro-lysimeters deviate by lysimeters. The three curves and bands in Fig. 6 are
BOAST & ROBERTSON: MICRO-LYSIMETER METHOD FOR DETERMINING EVAPORATION FROM BARE SOIL 693
-g 2
(a)
<D
0.0
Table 1—Mean and standard deviations of (a) time td, and (6) soil water loss, when cumulative evaporation from 70- and 44-mm
microlysimeters deviates by 0.5 mm from that of long microlysimeters.
(6) Water loss from soil in the long
(a) time t^ micro-lysimeters at time td
Initial Number 70-mm 44-mm 70-mm 44-mm
condition Evaporativity of runs micro-lysimeters micro-lysimeters micro-lysimeters micro-lysimeters
mm/d ———————————— d -
Wet 8.8 ± 0.8 3 1.4 ± 0.8 0.8 ± 0.5 7.7 ± 0.9 5.1 ± 1.4
Wet 5.5 ± 0.6 4 1.6 ± 0.4 (>!)* 0.7 ± 0.2 6.0 ± 0.7 3.7 ± 0.6
Wet 2.6 ± 0.3 5 2.9 ±0.8 (>!)**,(> 2)* 1.5 ±0.5(>1)* 7.6 ± 0.9 4.7 ± 1.1
Dry 6.2 ± 0.9 5 3.4 ± 1.3 (>!)**, (>2)* 2.5 ± 1.1 (>!)* 3.7 ± 0.7 3.0 ± 0.9
Dry 3.7 ± 1.0 5 4.0 ± 2.0(>1)* ,(>2)* 2.8 ± 1.3(>1)* 3.8 ± 0.7 2.9 ± 0.5
' One-tailed t-tests indicate whether td is > 1 or 2 d at the 95 (*) or 99% (**) confidence levels.
data are expressed as means and standard deviations wet and dry initial conditions, respectively.
over the groups. One-tailed Mests are used to deter- Strictly speaking, these relative errors are not com-
mine whether the 0.5-mm-deviation times (td) are sig- parable to each other because they occur after varying
nificantly > 1 or 2 d. Based on these /-tests and on amounts of time. However, most of the times are
Fig. 7, it is concluded that the 70-mm micro-lysimeters greater than 1 or 2 d, so the above errors are generally
deviate by < 0.5 mm from infinitely long behavior for larger than those which are encountered in the method.
at least (depending on initial wetness) 1 or 2 d, under The systematic error that occurs witrTmicro-lysim-
conditions where the evaporativity is 9 mm/d or less. eters can be estimated and corrected for by the fol-
The tendency, seen in Fig. 7a, for a deviation of lowing procedure, which is based on Fig. 6 and either
0.5 mm to occur more rapidly under conditions of high Table 1 or Fig. 77.mm
than low evaporativity is reflected in Table 1 by the 1. Estimate t d°' either using the upper regression
wet initial condition data, and, to a lesser degree, by line in Fig. 7a (for wet initial conditions, where
the dry initial condition data. The effect of the initial td is strongly dependent on the evaporativity) or
conditions is also shown in Table 1. In terms of real using Table 1 (for dry initial conditions, where
time, micro-lysimeters with dry initial conditions last it is not).
longer before a deviation of 0.5 mm develops than do 2. Divide the time that the micro-lysimeter has been
those with wet initial conditions. The opposite is true exposed to environmental conditions by f™~mm.
if the lifetime of the micro-lysimeters is expressed in 3. Find this value on the horizontal axis of Fig. 6
terms of the cumulative evaporation from the soil and estimate the correction (in mm), which is to
(right part of Table 1). be added to the measured value, from the middle
The two runs which were done on the samples with or lower curve of Fig. 6. (Note that for bothmm70-
a sharp increase in water content at about the 50-mm and 44-mm micro-lysimeters, the value ?™~ is
depth (filled circles in Fig. 2) are included in the av- used in this procedure in the denominator of the
erages in the bottom two lines of Table 1. It is inter- fraction tit™'""" because it is this value which was
esting to note that, for three of the four times and for used in developing Fig. 6.)
three of the four evaporation amounts given in these As a substitute for step (3) above, the following equa-
lines, the runs with frozen soil had values which were tions, which approximate the middle and lower curves
below the average values. Hence there is some in- of Fig. 6, can be used:
dication that the micro-lysimeter method is sensitive correction (in
to the unusual initial condition existing for these runs. mm) for a 70-mm
Error Analysis micro-lysimeter
If the absolute amount of error is important (for 0.0 if 0.0 < t/t7d°-mm < 0.5
example, if the total evaporation during an extended - 0.5 if 0.5 < t/ffmm < 1.0,
drying period is being determined by a series of micro-
lysimeter measurements) then the times between re- and
placement of soil in the micro-lysimeters can be longer
when the soil is dry than when it is wet. correction (in
If, instead, relative error is of concern, at least some mm) for a 44-mm
of this advantage for dry initial conditions is lost. For micro-lysimeter
initially wet 70-mm micro-lysimeters, the water lost
before a deviation of 0.5 mm occurs, regardless of the if 0.0 < t/t7d°-mm < 0.2
evaporativity, is about 7 mm (Table 1). Thus the rel- 0.2 if 0.2 < t/t™-mm < 0.5
ative error is about 0.5 mm/7 mm = 1% (after 1.4, 7 mm
0.7 if Q.5 < t/t d°- < 1.0.
1.6, or 2.9 d depending on the evaporativity). On the
other hand, for the runs with dry initial condition the The value of the proposed procedure for error es-
relative error is 13% (after 3.4 or 4.0 d depending on timation is restricted by limitations on the data from
the evaporativity). The corresponding relative errors which it is developed. It applies to only two micro-
for 44-mm micro-lysimeters are 11 and 17% for the lysimeter lengths and two levels of initial soil wetness
BOAST & ROBERTSON: MICRO-LYSIMETER METHOD FOR DETERMINING EVAPORATION FROM BARE SOIL 695
and is based solely on data from a fine-textured, high water content (gm H20/gm dry soil)
organic matter, surface soil. Refinement of the cor- 0.15 0.25
rection procedure and determination of the applica-
bility of the micro-lysimeter method to other soils
requires further work.
On the other hand, for the soil studied the error
estimation procedure is probably adequate for most so- -
uses. It provides an upper limit for the error because
the largest errors occur with wet soil and the micro-
lysimeter method seldom will be used under condi- runs 21 and 22
tions wetter than the wet initial condition studied. T) 100- - • Initial condition
(There is not much need to use the method in wetter A 146-mm m l c r o - l y s l m e t e r s
soil because soil evaporation is generally equal to or V 106-mm mlcro-lysimeteri
proportional to the evaporative demand when the soil O 70-mm mlcro-lyslmeten
surface is wetter than it was for the wet initial con- 15O- •
D 44-mm micro-lyslmelen
dition studied.)
Furthermore, in the range of soil moisture between Fig. 8—Final soil water content for two of the medium evaporativity
(5.4 and 6.0 mm/d) wet initial condition runs.
the two conditions formmwhich data are available, even
a poor estimate of ^°- leads to relatively small errors It is of further interest to note that for the other two
in the error estimate. Consider a one-day measure- wet initial condition, medium evaporativity runs, which
ment period with a 70-mm micro-lysimeter, under con- were terminated after only 1.9 d (not shown in Fig.
ditions of high evaporativity (9 mm/d) and of soil initial 8), the water contents of the 106- and 146-mm micro-
condition somewhere between wet and dry. At one lysimeters are very close to each other at all depths,
extreme, if the initial condition is assumed to be wet, an even stronger indication that these two sets of
from Fig. 7 or Table 1 $"-mm is 1.4 d, t/f^mm is 0.7, micro-lysimeters were still behaving as if they were
and the correction is 0.2 mm. If the initial condition infinitely long.
is assumed to be dry, from Table 1 f7mm is 3.4 d,
070-mm is QJ ; and the correction js Q.O mm. If the
measured soil evaporation is at least 2 mm (a reason- CONCLUSIONS
able expectation if soil which is not clearly dry is A new micro-lysimeter method can be used for di-
subjected to an evaporativity of 9 mm/d), the maxi- rect measurement of water evaporation from bare soil.
mum possible error in using the wrong correction For a mollisol and for evaporativity ranging from 2
value is 0.2 mm/2 mm (10%). Since in most cases an to 9 mm/d, micro-lysimeters 70 mm in length are ac-
intelligent guess can be made about whether to assume curate to within 0.5 mm for 1 or 2 d depending on
a wet or a dry initial condition, an error as large as initial soil wetness. A procedure for correction of sys-
10% will seldom occur. In a similar example, but with tematic deviations (due to the shallow no-flow bound-
low evaporativity, the correction for either wet or dry ary condition imposed at the bottom of the micro-
soil is zero. lysimeter) reduces the error further.
Finally, assumption of a dry initial condition for soil The method, while labor intensive, requires very
which is drier than that of this experiment also results little equipment. Evaporation can be determined under
in errors which are < 10% because the large ?™~mm circumstances where traditional methods are impract-
values associated with dry soil lead to small or no ical or impossible. For example, micro-lysimeters can
corrections. be used at a large number of locations for just a few
days, where the cost of larger lysimeters would be
Final Water Content Profiles prohibitive. Also, evaporation can be measured as a
A phenomenon which contributes to the ability of function of distance from a crop row, under conditions
the short micro-lysimeters to behave like infinitely of partial cover and partial shading, or in other situ-
long soil columns is shown in Fig. 8. Here gravi- ations for which the spatial resolution of traditional
metrically determined soil water contents of micro- lysimeters is too large.
lysimeters at the end of two of the wet initial con-
dition, medium evaporativity runs are plotted. These
two runs were terminated after 3.5 d, that is, consid-
erably after the 0.5-mm-deviation times: 1.4 and 2.0
d for the 70-mm micro-lysimeters, and 0.4 and 0.6 d
for the 44-mm ones.
The striking feature shown in Fig. 8 is that, although
the short micro-lysimeters lost considerably less total
water than the long ones during the runs, their soil
water contents decreased more. Note, however, that
the water contents in the top 20 mm of the 106- and
146-mm micro-lysimeters are nearly equal, a fact
which is consistent with the observed equality of evap-
oration rate for these two sets of micro-lysimeters
after 3.5 d.
696 SOIL SCI. SOC. AM. ]., VOL. 46, 1982