Beruflich Dokumente
Kultur Dokumente
AGRICULTURAL
AND
FOREST
METEOROLOGY
Agricultural and Forest Meteorology 88 (1997) 47-56
*, Per-Erik
Jansson
Department qf Soil Sciences, Swedish University of Agricultural Sciences, P.O. Box 7014, Vppsala S-75007, Sweden
Received 30 December 1996; received in revised form 20 June 1997; accepted 23 June 1997
Abstract
A modelling approach for predicting soil surface temperature and soil evaporation is presented. The procedure is based
on the equations for heat flow at the soil surface and includes vapour diffusion and a semi-empirical correction function for
the surface vapour pressure. The effects of changes in three important model parameters were studied by means of multiple
model simulations. The first parameter determines the steepness of the water potential gradient close to the surface. The
second parameter is the water vapour enhancement factor and the third one limits the lowest possible hydraulic conductivity
during drying. Measurements of soil water content and soil temperature in a bare sandy loam were used to evaluate the
models behaviour. Vapour pressure at the soil surface was found to be substantially lower than saturated vapour pressure at
topsoil moisture potentials as high as - 100 hFa. The difficulty to distinguish between vapour and liquid water flow at low
moisture contents was demonstrated. Results from the temperature tests indicated enhanced vapour diffusion and a probable
value of the diffusion tortuosity coefficient close to 1.0, whereas a value close to 0.7 was more likely according to the soil
water contents and calculated evaporation. 0 1997 Elsevier Science B.V.
Keywords: Energy balance; Soil surface; Vapour pressure; Soil temperature; Soil water content; Model; Simulation; Soil evaporation;
Vapour flow
1. Introduction
Evaporation from a bare soil surface is a complex
process including multi-phase transport of soil water
to the evaporating surface. This results in a vertical
redistribution
caused by simultaneous
liquid and
vapour water flows. In addition, evaporation depends
not only on soil properties but also on atmospheric
48
For a saturated
E,
can be written as
(2)
(3)
where (Y and p, often called the soil moisture
availability factors, are functions of soil water content that compensate for the changes in soil surface
humidity during drying. Comprehensive overview of
(Y and p formulations occurring in the literature can
be found in Mafouf and Noilhan (1991), Lee and
Pielke (1992), Mihalovid et al. (1993), and Deki6 et
al. (1995).
From thermodynamic
laws, Philip (1957) derived
an expression for the relative humidity of air in
equilibrium with the water in the soil pore, h, as:
h=exp($M,g/R(T+273.15))
(4)
2. Materials
and methods
as the
of heat
caused
and v
49
(11)
where q is flow, k is conductivity, T is the temperature, z is depth, c, is the specific heat of water and
A is the heat of vaporization.
Soil water flow is comprised of saturated flow
according to Darcys law as generalized for unsaturated flow by Richards (1931) and water vapor flow,
q,, as follows:
4, = k
d+
-z+1
+4v
(6)
dv,D,fa
cs- csoil
(12)
AZ
2
(13)
R,=hE+H+G
H= pacp
(14)
(8)
(K-T,)
ra
1)
(15)
50
face could decrease by up to two powers of magnitude during drying and increase by up to one power
of magnitude during wetting.
The precise predictions of soil surface evaporation
will thus include empirical parameter values for
vapour diffusion in the soil, d,, and the non-equilibrium coefficient, $s. In reality, the unsaturated
conductivity, k,, is also difficult to estimate under
dry conditions. To investigate the importance of the
unsaturated conductivity values at low water contents the k, equation by Mualem (1976) is limited
with a minimum value of unsaturated conductivity,
k,. Thus,
k, = max[ks( f)2@l*.
k-1
(16)
ously. Thin thermocouples (type T, copper-constantan) for soil temperature measurements were installed at l-cm depth intervals from 1 to 6 cm depth
in two profiles in the plot, while temperatures at
deeper levels (10, 20 and 30 cm) were measured
only in one profile. Soil water contents were measured with Time-Domain Reflectometry (TDR)
technique in two similar profiles at 2, 4, 6, 10, 20
and 30 cm depth, respectively. Miniature tensiometers were used for soil water tension measurements
at the same levels as the TDR probes in one profile.
Soil surface temperature was determined both with
four thin thermocouples and with an infrared thermometer scanning over the plot. Soil surface moisture was measured with a gas analyser on air sampled close to the soil surface. The plot was equipped
with a miniature net radiometer.
2.3. Soil properties
Soil texture from 0 to 30 cm depth was determined at 5-cm depth intervals for lumped soil samples at three different spots. The soil profile from 0
to 30 cm depth was homogeneous, consisting of
approximately 10% of clay, 13% of silt and 73% of
sand and with an organic matter content of 3-4%.
The bulk density increased from 1.40 g cme3 at O-5
cm depth to 1.63 g cme3 at 25-30 cm depth.
Small soil cores (5 cm high, 7 cm of diameter)
were sampled from the surface down to 30-cm depth,
with three replicates at each level, to obtain informa-
Fig. 1. (a) Soil water retention curves for depth intervals of O-5 and 5-30 cm. The curves are fitted to measured values (crosses and
squares) using the Brooks and Corey expression. (b) Unsaturated conductivity for O-30 cm depth calculated according to Mualem (lines)
and calculated from measurements (crosses = 3 cm, filled circles = 10 cm and squares = 20 cm depth).
G. Aluenti,
P-E. Jansson/Agricultural
40
51
SOlI 5 em depth
(C)
30 20-
ii.
i,
9
10 -
0,
30 _
I-
Scmdepth
(%)
1:
12
The SOIL model (Jansson, 1991) was parameterized for 14 layers, the thickness of which increased
slowly with depth, starting with 2 cm layers in the
uppermost 8 cm of the profile. The meteorological
station at the site provided data for the driving
variables. Measured temperature and water content
were used as initial values. To represent a plain bare
soil surface the surface roughness was estimated to
be 0.001 m.
Three model parameters affecting soil surface
moisture and temperature were selected for a sensitivity test, namely the I,!+, k, and d,, parameters.
The first parameter governs the surface moisture
under both dry and wet conditions, while the importance of the two latter increases when the soil becomes dry. The I& parameter was increased in steps
of 0.4 from 0.0 to 2.0 which covers a wide range of
moisture gradients in the uppermost layer, allowing
the surface water availability to decrease up to lo4
times during drying. A commonly used value of the
d,, parameter in the diffusion equation is 0.66, but
higher possible levels, even above 1.0, have been
suggested (Philip and de Vries, 1957; Cary, 1963;
Cass et al., 1984). This enhanced vapour flow may
be caused by e.g., local temperature gradients or
evaporation/condensation
on different sides of thin
water films in the soil (Philip and de Vries, 1957).
An extensive overview over different explanations of
the enhancement phenomenon and a presentation of
vapor diffusion enhancement
factors occurring in
literature can be found in ten Berge (1986). In the
present test, the d,, parameter was increased from
0.1 to 1.0 in steps of 0.3 to cover both low-diffusion
scenarios and enhanced vapour diffusion. The minimun hydraulic conductivity,
k,, was increased tenfold in every step :from 10- to 10 mm/day,
Temperature
88 (1997) 47-56
27
30,
June
Fig. 2. Measured air and soil temperature
5-cm depth.
J&
and soil water content at
52
6
3
3cm
0.6
lcm
7
2
Scm
i
5
10
3
IO
20 cm
cm
cm
20 cm
-10
0
-10
-4
k,
(Iog2mm,d:y)
-4
km
3cm
-4
-4
(log mm/day)
at
Scm
Root
10 cm
mean square
errors
(mm)
1.0
6
20 cm
0
-10
0
-10
4
k,
2
(log
-4
-4
km
mm/day)
(kw
mm/&y)
Fig. 6. Root mean square errors for calculated soil water losses
and simulated evaporation (explanations are the same as for Fig.
3).
53
54
G. Alum&,
ment uncertainties mentioned earlier and to the uncertainties in the k(JI) relation as discussed below,
the k, interpretation is afflicted with uncertainties.
Above lo- mm/day
the coefficient of determination dropped dramatically to approach 0 with further
increased
k, values irrespective of the diffusion
rate. This indicates that such high minimum conductivities could not be used.
When evaluating the k, parameter it should be
kept in mind that the relation between unsaturated
conductivity and soil water content in the dry range
is uncertain. Because direct measurements
are time
consuming, expensive and difficult to perform (e.g.,
Ragab et al., 1981, or van Genuchten et al., 1989)
the k( I/J) relation in this case was calculated from
measured saturated conductivity, k,, using the equation of Mualem (1976). However, as pointed out by
Khaleel et al. (1995) among others, using k, as a
single matchpoint for the k($) curve may result in
inadequate characterization of the relationship in the
dry moisture range. Although it is true that the k(JI)
curve was also compared with conductivities calculated from water contents and tensions measured
during a drying period (Fig. lb), the scatter in the
measured-calculated
values and the narrow range
within which the tensiometers can be used leaves
room for different interpretations
of the k( I,!I) relation. Slightly changing the shape of the kc+) function might give a result similar to that obtained by
changing the k, value. Methods using soil evaporation as a boundary condition for calculating k( I/I> in
the dry range, e.g., the hot-air method (Arya et al.,
19751, naturally cannot be used as validation since
they include an unknown contribution
of vapour
flow.
3.3, Soil evaporation
As for the water contents, no clear indications of
enhanced vapour diffusion were given by comparisons between measured soil water losses at O-30 cm
depth and simulated soil evaporation (Fig. 6). At
high diffusion rates (dvb = 1.0) the vapour flow to
the surface caused an overestimation
of the simulated evaporation
irrespective of the level of the
lowest possible hydraulic conductivity (k, >. Similarly, low diffusion rates (dvb = 0.4 or 0.1) resulted
in underestimated
simulated evaporation owing to
88 (1997) 47-56
55
the limited amounts of moisture that could be transported to the surface. With increasing minimum
hydraulic conductivity ( k, >, however, the moisture
deficiency at the surface turned into a surplus, leading to overestimated evaporation, particularly when
small moisture gradients close to the surface (low I,$
values) were presumed to exist.
The dependence of simulated evaporation on the
slope of the near-surface moisture gradient was
clearly demonstrated in this study. A good agreement
between simulated e,vaporation and water losses calculated from measured soil water contents could
only be obtained when a I,$ value of 1.2 or higher
was used. This result restricts the possible range of
I,$ values more strongly than restrictions obtained
with the temperature and water content variables,
thus implying that rsurface vapour pressure already
starts to decrease at surface moisture potentials as
high as - 100 hPa.
As for the temperatures, minimum conductivity
levels above 10-5-10-4 mm/day resulted in poor
agreement between simulated and measured water
losses.
4. Conclusions
Due to the higher accuracy of temperature measurements compared with soil water measurements,
soil temperature was more useful than water content
for evaluating methods of estimating soil surface
energy balance. However, evaporation calculated
from changes in the soil water storage was a useful
complement that could be used to restrict the range
of some crucial and uncertain properties of the system.
Soil temperature, soil water content and evaporation all indicated that the use of the correction factor,
s, is necessary to prevent soil evaporation rates
CF,
from being too high. +s values as high as 1.2 had to
be used for the soil in this study. This level is in
agreement with the results presented by Schelde et
al. (1997).
Good agreements between measured and simulated temperatures were achieved when d, values
close to 1.0 were used. This result supports the
theories of vapour flow enhancement proposed by
Philip and de Vries. However, the soil water content
Acknowledgements
References
56
88 (1997) 47-56