Journal 10
Journal 10
Journal 10
ABSTRACT
In large-scale models, an accurate calculation of energy fluxes at the lower boundary is
important to ensure satisfactory predictions of the meteorological parameters. This work is
performed using surface models.
In this paper. various surface models. previously studied by Deardorff are tested against
experimental data from the West African Monsoon Experiment 1979 in the Ivory Coast.
Physical properties of the soil are assumed constant and uniform. Input variables are dry and wet
bulb temperatures. wind velocity at the 2 m level, and global radiation. We compute sensible and
latent heat fluxes and the surface conduction flux. A Crank-Nicholson scheme has been used
with a time step of 5 min. Estimated sensible and latent heat fluxes agree well with experimental
data. The order of magnitude of the difference between theoretical and experimental values is
30 W m-*forevaporation and 20 W m-* for sensible heat. Surface soil heat flux ie less satisfactorily
calculated. which could result from properties assumed constant such as soil surface humidity. A
model including the heat conduction in the soil provides the best results. The others, based on
empirical formulations of surface soil heat flux, are less satisfactory. However. with a simple
two-layer model. the predictions are quite acceptable.
is a field with a fetch of some hundred metres (a 5"30'W, about 400 m above sea level, in the
single site of data collection) but it should be noted Northern Ivory Coast) during WAMEX (West
that such measurements are very rare in the African Monsoon Experiment) in 1979 by the
African savannah. Laboratoire Associk de Meteorologie Physique
Experimental data have been collected on a (L.A.M.P., Universite de Clermont-Ferrand 11,
sub-sahelian savannah site (Korhogo, 9O30' N, France). The experimental area was partly covered
with grass (about 30 cm high). The experimental
device consisted of a 6 m high mast with three
Table I. Rainfall (mm)recorded from July 1 to
measurement levels: 50 cm, 2 m and 6 m. The
August 10 at Korhogo during WAMEX 1979
~ ~~ ~~
following data were recorded: wind speed (cup
anemometers), temperature and humidity (dry and
July July July August
wet bulb ventilated thermometers), wind direction
(at the top of the mast). Moreover, six thermo-
day rainfall day rainfall day rainfall day raintall couple gauges measured the soil temperatures at
~
Fig. I . Meteorological parameters during the period July 31 to August 10 (1979) at Korhogo: R, = global
radiation; Ta = air temperature at 2 m above ground: qs and u, = respectively specific humidity and wind speed at the
same level (2 m).
heat flux becomes quite negligible at a depth of 80 kept constant. A test prohibits qo from becoming
cm, the surface value of this flux, Go, is given by the greater than qSat(t9,)when condensation occurs (at
integral: that time, a equals 1).
The infrared budget is given by the Angstrom
formula:
G, = C[*$ dz (z2 = 80 cm), (1)
R,, = f ( v ) (RT - &, RL), (5)
where 8, is the experimental soil temperature profile
where
and C the specific heat of soil, assumed constant
and uniform. A continuous set of 11 days has been R, = &, ae;, (6)
used in this work. The soil parameters such as
thermal conductivity, specific heat and wetness, R, = (0.82 - 0.25 * 1 0 ~ 0 . 0 5 2 e a ) uT:. (7)
and surface properties such as albedo and emiss- f ( v ) accounts for the effect of cloudiness v. It is
ivity were not directly measured. The investigated assumed that
period (July 31 to August 10, 1979) took place in
the middle of the rainy season. This period is f ( u ) = 1 - 0.55v (8)
characterized by an alternation of rainy days (Budyko, 1974; Reed and Halpern, 1975;
(squall lines on July 31 and August 5) and sunny Coulibaly, 198 1). A rough estimate of cloudiness is
days. Wind speeds are generally very low. Aside derived from the formula:
from occasional disturbances, they are lower than 4
m s-I. Values ranging between 2 and 3 m s-' are R, = RGO(1 - C,4. (9)
frequently observed. During daytime, instability is
from which v is deduced:
strong: the Richardson number may be as low as
-0.5. Rainfalls from July 1 to August 10 are V= (RGo - R,)/C, RGO, (10)
presented in Table 1. Fig. 1 summarizes the main
where R,, is the global radiation when the sky is
meteorological features from July 3 1 to August 10.
clear. This value is estimated by astronomic
A detailed description of this period has been
formulas (Paltridge and Platt, 1978), with an
presented by Coulibaly (1981).
atmospheric attenuation of 20%. Adjustment runs
lead to a value of 0.8 for C,.
2. Description of the models 2.2. The soil
2.1. The atmospheric surface layer 2.2.1. The multi-layer model. In a medium with
uniform physical properties, the heat conduction
The description below is common to all models.
equation is
In the lowest part of the atmosphere (where the
fluxes are assumed conservative) the various terms a8(z,f)
~-
a2O(Z, t)
of the heat budget equation (except Go) may be -K-.
at aZ2
expressed by bulk formulas:
The lower boundary condition is 8 = 8,. This
H = P. c, c,, u,(eo - T.) (24 parameter is equal to the value of the temperature
LE = LPn chO Ua(qO - 4a) (2b) at 80 cm depth. For a 24-h period, it remains
practically constant. At the soil-atmosphere inter-
q,, is interpolated from qsUt(8,)and qa in the follow- face, the heat budget equation, which accounts for
ing manner: no heat storage in a thin film at soil surface,
provides the upper limit condition for the diffusion
40 = a s , , t ( ~ o ) + (1 - a) 4.9 (3)
equation:
so that eq. (2b) becomes:
H + LE + R I R - ( 1 - A ) R, + Go = 0. (12)
LE = 0,ChO u, a(ssnt(e0)
- 4,). (4)
The surface soil heat flux by conduction is found
Since the models do not take the soil water from
budget into account, the parameter a which
characterizes the soil wetness near the surface is Go = -,I(a@/az) at z = 0. (13)
Tellus 37B (1985), 2
CALCULATION OF GROUND TEMPERATURE AND FLUXES 61
The input data for the model are: of the ground and an inner level (at depth d , ) at
the wind velocity at 2 m, which the temperature is em. Eqs. (16a) to (16e)
the wet and dry bulb temperatures at 2 m, are linearized, using for aO,,/at the analog (&,,+l -
the global radiation (surface value). &,,,)/At (the subscript n + 1 refers to the present,
Eq. (1 1) is discretized following the Crank- unknown time step, and n refers to the past time
Nicholson scheme. There are ten levels in the step). All the terms of the heat budget equation
one-dimensional grid used in the integration: these (except G,,) are expressed in the same way as for
levels correspond to depths expanding exponen- the multi-layer model. The input data are the same
tially (except the last): 0 (surface), 6 mm, 12 mm, as for the multi-layer model.
25 mm, 5 cm, 10 cm, 20 cm, 40 cm, 80 cm, 1 m.
The time step Af is constant and equal to 5 min.
Tests on At, allowed by the implicit character of the
scheme, showed negligible truncation errors (a run 3. Initial choice of the basic parameters
with At = 30 s was performed with results very
close to those obtained with Af = 5 min). The The basic parameters A, K, 1, a, C,, and E , have
non-linear terms are linearized as follows: been chosen by adjustments. An initial choice of
these parameters is performed as follows:
qSatlO(t+ At)] = qsat[B(t)l+ lO(f + A 0 - OW1
albedo A : 0.20 (Rockwood and Cox, 1978).
K and 1:in agreement with measurements in the
laterite, a value for K of 0.7 mz s-I was taken
e4(t + At) = 48(f) @(I + Af) -384(f). (15) (Baudet and Degiovanni, personal communica-
tion). An estimate of 1is derived from the specific
2.2.2. The empirical models. The five empirical heat C taken as 2.2. lo6 J kg-I K-by Coulibaly
models previously tested by Deardorff are evalu- for the processing of the WAMEX data used here.
ated here by direct comparison with the ex- From these values of K and C, an initial value of 1.5
perimental data. They are based upon solving the W m-I K-I has been selected for 1.These values
heat budget equation (12). Go is expressed in agree with De Vries data (1966) for wet soils.
various ways, which distinguishes one model from a = 0.5 (the soil was very wet).
another. According to the expression chosen, they C,, = 5 . lo-. This value agrees with a rough
are referenced as 1, 2, 3, 4 or 5 . The different estimate from the formula:
formulas for G, are described thereafter:
C,, = u* T*/(Ua(eo - Ta)) (17)
model 1 :Go = 0; (164
(u, and T , are the scaling surface parameters for
model 2: G, =fH ( 16b) momentum and sensible heat). Typical values of
u,, T,, u, and 0, - T , are, respectively, 0.3 m SKI,
model 3: 0.3 K, 2 m s- and 10 K , so that C,, is about
5 . 1 0-. Formula (1 7) is obtained if we compare the
G,, = -0.19RN when R, < 0 (day) expression of H from the aerodynamic method, i.e.
( G,, = -0.32RN when R, > 0 (night);
(164
pa C, T, u, (Saugier and Ripley, 1978) with the
bulk formula (2a). (This estimate of Choby formula
(17) is rather crude because no measurements of H
model 4: G, = n-12Cd,(af?,af) by some other method than the aerodynamic were
(forcing process); (16 4 available).
In order to study the effects of changing these
model 5 : G, = 0.5n-12Cd1(&9,lar) values, sensitivity runs have been performed for a
typical day (August 7). Some examples are given in
Figs. 2 and 3, for the maximum values of the
surface temperature On issued from the multi-layer
This last model, called by Deardorff force-restore model, referenced at c( MLM, and from model
model, is characterized by two layers: the surface 5 (two-layer model), referenced as T Z ; 2LM. The
Tellus 37B (1985),2
68 G . CAUTENET, Y. COULIBALY AND Ch. BOUTIN
r--
L5t
Fig. 2. Sensitivity of surface temperature (extreme values) estimated by the multi-layer model (symbol MLM) and by
the two-layer model (symbol 2LM) and fluxes H, LE estimated by the multi-layer model (maximum values) to
humidity parameter a and thermal inertia I (August 7, 1979).
1 3
f '".
I 1
I .
U> gri
ALEE00
Fig.3. Sensitivity of surface temperature (extreme values) estimated by the multi-layer model (symbol MLM) and by
the two-layer model (symbol 2LM) and fluxes H , LE estimated by the multi-layer model (maximum values) to albedo
A and bulk coefficient Cho.
Tellus 37B (1985), 2
CALCULATION OF G R O U N D TEMPERATURE A N D FLUXES 69
corresponding minimum values are also repre- surface temperatures are not affected by the initial
sented. Moreover, we present the maximum values conditions, even for departures of several degrees in
of fluxes H and LE issued from the multi-layer the initial temperature. After interpolation, the
model. For a variation of the parameters around measured temperature profile at 0 h (local time) is
their initial values, an increase of 1 K in ys used to initialize the multi-layer model. As there is
MLM is expected when: experimental evidence that at 80 cm depth the
A decreases by 0.1, or: temperature gradient is weak and the temperature
the thermal inertia I (equal to LK "*) is reduced itself varies very slowly, 0, is assigned this experi-
by 400 W m-2 kk's''', or: mental value of 0 h. Moreover, 8, is not allowed to
a decreases by 0.1, or: vary for a given day. (Experimentally, the changes
C,,, decreases by 10~'. did not exceed 0.1 K per day.) The same value of
As regards the surface emissivity E, (results not 0, is used in the multi-layer model and in model 5.
presented here), there was little sensitivity in the The multi-layer is not sensitive to variations of
range of its possible values: the decrease is about initial temperature in the upper soil layer
0.75 K when E, is increased from 0.80 to 1. The (0-20 cm).
maximum TEr; 2LM varies approximately in the
same way. 4.2. Adjustmenr ofthe basic parameters
The latent heat flux seems more sensitive than This adjustment was performed using the multi-
the sensible heat flux. The responses of the layer model. The parameters, one at a time. were
maximum values of H and LE to changes in I , A modified so as to get a good fitting of the
and C,,, are similar: they decrease when I and A experimental values of the soil temperatures and of
increase, or when C,, decreases. On the other hand. the fluxes by the multi-layer model. The modi-
when the parameter a increases. H decreases but fication of the basic parameters was performed
LE increases, because the effect of a greater with the help of the results of sensitivity runs. This
moisture at the surface overcomes the moderating adjustment may be described briefly as follows: for
effect of the decrease in the surface temperature. every day. the thermal inertia I was slightly
The lowest soil layer temperature 0, was also modified and the model was run in order to get the
investigated. It moderately influences Ty::;",; 2LM: theoretical curve 0, close to the experimental one.
an increase of about 0.25 K is observed when 8, is The C,,, was adjusted so as to get a good agreement
increased by 1 K. On the other hand. in the same for H . Finally. the wetness a was modified so as to
conditions. the multi-layer model is almost in- get an acceptable curve for LE. At every step, if
sensitive to 19, for a 24-h run. When this value is necessary, all the parameters already estimated in
not experimentally available. Blackadar ( 1976) the previous steps could be slightly modified in
stated that it could be estimated from the mean air order to preserve the fitting already fulfilled. I f one
temperature of the previous day. The last test of those steps could not be achieved, another value
concerned the effect of changes in K and 1 with of albedo A was chosen and the complete cycle was
constant I : only the inner layer temperatures are re-run so as to get a good general agreement. The
changed, but not the surface temperature. which is last step consisted in the choice of the best value of
governed by thermal inertia. This is in agreement K (or A): for the value of I found in the first step, K
with the results of Carlson et al. (1981). was modified (and 1) in order to get good
agreement in the ground temperatures at the 10 cm
and 20 cm levels. Due to the lack of sensitivity to
this parameter. the emissivity E , was chosen equal
4. Verification of the models to 0.9. All the test runs showed that the best value
for C,, was 5 . 10 3. The hypothesis of constant C,,,,
4 . 1 . Initialization ofthe models is questionable, because this bulk coefficient is
Integration begins at 0 h (local time). For the likely to be an increasing function of the instability
empirical models, the initial surface temperature in the atmospheric surface layer (Greenhut, 1982).
was chosen as the air temperature at the same Nevertheless. the behaviour of C,, associated with
hour. Great accuracy in this initial value is not weak wind speeds is not yet completely known. and
required: after 2 h of simulated time, the calculated for the purpose of the present study. a constant
Tellus 37B (1985). 2
I0 G . CAUTENET, Y. COULIBALY AND Ch. BOUTIN
2 4 6 8 10 12 li '6 18 20 2 2 24 2 L 6 8 10 12 1L 16 16 20 22 24
I ' I ' I I /
600 -
-
t 300
5
21n
CALCULATION OF GROUND TEMPERATURE AND FLUXES 71
2 4 6 a 10 12 v. 16 ia m 22 21 2 1 6 8 10 12 11 16 18 B 22 )L
Local time Local tlme
10) It)
-
X 2 L
L
6 8 10 12 14 16 8 20 22 21
d -200
l
2
!
i
l ! i l i l l l l -
6 9 I0 12 11 16 18 20 22 16
-1
Locol time Local time
I bl Id1
these runs range between the following limits: 4.4. Temperatures from the multi-layer model
1.50 .: 1 < 1.60 (W m-I K-I); The surface temperatures issued from the multi-
layer model are in rather good agreement with the
0.65. < K < 0.80 (m2s-I); experimental data (Figs. 4a, 5a and 6a). It may be
noticed, however, that the nocturnal values are less
0.15 <: A < 0.30; satisfactory, probably because the bulk coefficient
0.40 <: a < 0.70. is overestimated by night: a lower value of C,,
would lead to lower absolute values of fluxes H and
It may be noted that the range of the possible LE, which are negative by night. As the surface
values for K and 1is not very large. temperature governs sensible and latent heat
exchanges, and these fluxes are very weak by night,
accuracy in determining the surface temperature is
4.3. Presentation of the result set much more necessary by day, when H and LE are
Figs. 4-6 present examples of calculations high, than by night. The errors d x y range between
compared with experimental data for some typical 1.05 and 2.5 K, with an average value of about 2
days. On July 31, a squall line occurred in the K. The correlation coefficients are satisfactory,
morning (Fig. 4); August 2 and 10 were sunny except for July 31. On this day there was rainfall
(Figs. 5 and 6). Table 2 summarizes the statistical (24 mm) in the morning at 10 h (local time), but the
parameters for the multi-layer model: correlation model as presented here does not take rainfall into
and regression coefficients and distances between account. In Fig. 4a, the experimental curves, but
experimental and calculated values (i.e. the model not the theoretical curves, reveal an obvious
errors) for the surface temperature, H and LE. cooling (particularly at 10 cm).
Tellus 37B (1985), 2
12 G. CAUTENET, Y. COULIBALY A N D Ch. BOUTIN
- 45
-
LO
CL
Y
2 35
9
eT 30
25
20
-2 M
10 1 1 1 1 1 1 1 1 I l 1
1 2 1 6 8 10 512 11 16 18 21k 2 1 6 6 10 12 IL 16 I8 10 21 ZL
Locol lime Locol time
10 1 IC1
' 15
10 1 1 1 I 1 1 I I I 1 1 -zoo LuuuuL
2 L 6 8 10 12 1L 16 I8 20 22 24 2 4 6 8 10 12 1L 16 18 20 22 2L
Locol lime local lime
101 Id 1
Table 2. Statistical results from comparison of multi-layer model with experimental data (soil parameters
adjusted every day)
31-07 01-08 02-08 03-08 04-08 05-08 06-08 07-08 08-08 09-08 10-08
-e
a
a
b
1.12
-2.94
1.19
-5.02
0.94
0.98
1.03
-2.65
0.90
1.50
1.29
-8.27
1.13
-4.05
1.01
-1.34
1.06
-3.26
1.00
-1.68
0.97
-0.59
E g r 0.83 0.98 0.96 0.91 0.97 0.94 0.97 0.95 0.94 0.97 0.95
e= dxy 1.05 1.45 2.11 2.80 2.40 1.56 1.65 1.99 2.10 2.14 2.48
zg ax, 0.19 0.11 0.10 0.18 0.11 0.18 0.09 0.12 0.15 0.13 0.15
H a 1.26 1.13 0.89 0.98 0.88 1.11 0.80 0.85 0.61 0.86
b 14.1 20.3 16.9 14.4 17.4 18.4 m 12.6 7.80 15.1 13.3
r 0.84 0.85 0.98 0.93 0.95 0.89 4 0.95 0.98 0.93 0.96
dxy 19.0 28.4 18.5 22.6 18.9 21.3 Z 14.5 11.4 24.3 16.2
0.24 0.09 0.16 0.17 0.26 0.12 0.07 0.12 0.10
S, 0.28
z
L -
w
LE a 0.86 0.99 0.88 0.78 1.03 0.98 2 0.90 0.85 0.81 0.75
b 15.81 -1.57 6.52 8.67 2.29 19.9 ," 8.05 -0.25 5.87 4.92
r 0.83 0.97 0.98 0.98 0.93 0.94 2 0.94 0.99 0.97 0.97
dxy 23.9 22.1 29.8 31.9 25.5 27.2 -3 26.6 27.5 27.5 37.5
S, 0.17 0.08 0.09 0.08 0.14 0.1 1 0.1 1 0.07 0.07 0.1 1
As regards the surface temperature, it is im- flux seems more accurately calculated than the
portant to keep in mind that the experimental sensible heat flux. High-frequency fluctuations of
parameter considered to represent it, is merely the measured fluxes may be noticed in model issues
output of the thermal gauge at depth zero, i.e. lying also. The correlation coefficients are high, with
straight on the ground and covered with a fraction lower values on the days with squall lines, i.e. July
of millimeter of earth. Direct measurement of the 3 1 and August 5 .
exact boundary temperature-the so-called skin
temperature-is not possible. This parameter 4.6. Surface soil heatflux Go
results from various processes (conductive, con- It is obvious from Figs. 4d to 6d that the
vective and radiative), unlike the temperature in agreement between theoretical and experimental Go
any inner layer of the soil, which proceeds, neglect- values is less satisfactory than for H and LE.
ing water transfer in the ground, from conductive Particularly, a phase lag of the experimental curve
processes only. Radiometric measurements of with respect to the theoretical one may often be
surface temperature (from aircraft or satellite) observed. This lag is about 1-2 h. Depending on
would be more relevant on theoretical grounds, but the day, the correlation coefficient may vary
during this phase of the experiment, no accurate between the limits 0.65 and 0.83. We did not
available data exist. As a matter of fact, surface present here the experimental errors on Go,
temperature calculations by models take the afore- because, in the determination of this flux, the
mentioned processes into account. Thus, they are specific heat C was not measured directly (see
somewhat connected with radiometric tempera- Section 3): thus, the errors on the vertical profile of
tures. The calculations, however, agree with the C are not available. It is realistic, however, to state
experimental data. Temperatures at 10 cm and at that these experimental errors could not account
20 cm also closely agree. Theoretical and ex- thoroughly for the departures between experi-
perimental curves appear much smoother than mental and theoretical values: their somewhat
surface curves. This fact may be easily explained: systematic character as well as the shape of
the surface temperature may be represented, as it is theoretical curves suggests that they originate more
a quasi-periodic function of time, by a sum of pure from the model itself.
harmonic waves of increasing frequencies. The
higher frequencies contributions (originating, say, 4.7. Surface temperatures from the empirical
from rapid changes in cloudiness, see e.g. Fig. 6a) models
may be important. Now, the damping of a thermal These models use the same values of the basic
wave of frequency f with the vertical coordinate z is physical parameters as the multi-layer model. Figs.
proportional to exp (-Kf12 z), where K is a 4b to 6b show the surface temperature from the
constant depending only on the type of soil empirical models. Since it governs the fluxes H a n d
(Carslaw and Jaeger, 1978). Thus, the high-order LE, the accuracy of these models may be evalu-
harmonics are quickly attenuated with increasing z: ated by examination of this parameter. For this
the soil acts as a low-pass filter. This softening of reason, the fluxes H and LE calculated by the
the shape of curves is confirmed by experiment, empirical models are not presented here. All of
and the model accounts for it too. these models provide results markedly different
from the experimental data, except model 5 . This
4.5. Fluxes H and LE from the multi-layer model two-layer model provides issues surprisingly com-
On the curves of Figs. 4c to 6c, the nocturnal parable with those of the more complex multi-layer
values are generally omitted because they are weak model.
(about some W m-) and highly fluctuating, with
theoretical as well as experimental values (the
nocturnal values of wind velocities do not allow a 5. Two supplementary tests
high accuracy in flux determination by the aero-
dynamic method). Table 2 shows that the mean 5.1. A run with conslant basic parameters
values of the square distances (errors) for H and Table 3 refers to a run for which the basic
LE are 20 W m- and 28 W m-*, respectively, i.e. parameters K, 1, A and a (and, of course, E, and
16% and 10% in relative values. The latent heat C,,,,) were kept constant over the whole period of 11
Tellus 37B (1985), 2
74 G. CAUTENET, Y. COULIBALY AND Ch. BOUTIN
Table 3. Statistical results from comparison of multi-layer model with experimental data (soil parameters
hold constant over the whole period)
31-07 01-08 02-08 03-08 04-08 05-08 06-08 07-08 08-08 09-08 10-08
~ ~~ ~
a 0.92 1.21 0.85 0.96 0.96 1.42 1.19 1.06 1.06 1.02 0.95
E
w
b 0.15 -5.42 2.68 -1.72 -0.25 -12.8 -5.61 -3.44 -3.93 -2.72 -0.81
r 0.64 0.97 0.96 0.91 0.96 0.88 0.97 0.95 0.95 0.97 0.96
E S d,, 2.28 1.71 2.65 3.44 2.55 2.82 1.89 2.42 2.57 2.54 3.04
E n
3 g
lnnr
S, 0.42 0.13 0.13 0.22 0.11 0.32 0.11 0.14 0.18 0.15 0.18
H a 1.34 1.23 0.74 0.79 1.00 1.39 0.88 0.78 0.60 0.76
b 1.07 22.9 11.7 4.87 16.1 6.23 6.14 1.61 10.2 7.85
r 0.87 0.86 0.98 0.91 0.95 0.91 3 0.95 0.97 0.91 0.96
dxy 11.9 31.7 19.7 19.0 19.9 15.6 ';ii 11.0 13.0 24.4 15.1
S, 0.17 0.27 0.10 0.13 0.18 0.19 0.09 0.08 0.12 0.10
2
.-
LE a 0.82 0.90 0.88 0.78 1.45 1.00 3 0.95 0.76 0.80 0.81
b -10.7 -0.55 4.35 -2.77 -2.91 14.6 2 -3.04 -9.40 -2.77 0.14
r 0.81 0.97 0.98 0.97 0.93 0.94 2 0.94 0.99 0.96 0.97
dxy 25.3 24.6 30.2 39.2 50.4 23.3 -3 28.5 42.1 32.5 33.9
S, 0.18 0.09 0.09 0.09 0.28 0.10 0.11 0.11 0.09 0.10
a .?
E
3
-
1
0 1 " " " " " "
0 2 I 6 8 10 12 14 16 18 M 22 2L
local time
Fig. 7. Variation of a during daytime for the simulation
-200 I 1
of drying of the upper soil layers.
LOCOI tlme
days. The values chosen are 1 = 1.55 W m-I K-'; Fig. 8. A simulation of soil surface drying during
K = 0.7 m2 s-I; A = 0.20; a = 0.40, which are daytime. Effect on Go (experimental Go:-; calculated
with constant a: ----; with variable a -.---)
crudely the mean values of those relative to the (August, 2).
previous case. This run shows slightly more
important deviations, but the correlation coeffi-
cients remain the same. This trial simulates the use sumption that the soil properties are constant. A
of surface models with seasonal values of the run with a taken as a function of time has been
physical parameters. performed. Drying of the upper soil layers during
the first part of the day is simulated by assigning to
5.2. Use of a variable wetness parameter a a decreasing value from 10 a.m. to 2 p.m. (local
In Subsection 4.6 are pointed out some dis- time). Afterwards, a begins to grow slowly again
crepancies between theoretical and experimental (which accounts for wetness restoration in the
values of Go. These may result from the as- upper layers by capillarity from the lower layers).
Tellus 37B (1985), 2
CALCULATION OF GROUND TEMPERATURE AND FLUXES 15
.oo r
- r of H and LE by the multi-layer model are only a
small % greater than in the case of the daily
C, specific heat of air at constant pressure. T,,,,[, TI, and T I , temperatures at depths 0, 10
d, parameter representing a n order of mag- and 2 0 cm (symbols used in Figs. 4 t o 6).
TmaX
nitude of soil thickness under influence of the rurf MLM maximum value of the surface tem-
diurnal temperature variation, d, = ( ~ 5 ) ~ . perature (evaluated with the multi-layer
e, mean value (at daytime scale) of e, (air model).
vapour pressure). TT;: 2LM maximum value of the surface tem-
esat(T ) saturation vapour pressure at tempera- perature (evaluated with the two-layer
ture T. model).
G, surface soil heat flux. f time.
H sensible heat flux (surface value). At time step in discretized equations.
I soil thermal inertia ( I = L K - I ~ ) . u, wind speed at level 2 m.
L latent heat of evaporation for water. z vertical coordinate (positive upwards; origin
LE latent heat flux (surface value). 1 m depth).
9a specific humidity at level 2 m. a adimensional humidity parameter for soil
90 specific humidity for air at ground surface. surface (lying between 0 and I).
9sal(T ) saturation specific humidity at tempera- F, surface emissivity.
ture T. 0 ( z , t ) thermal profile in the soil (theoretical
net IR budget at ground surface. values).
global radiation at surface level. 0,(z, t ) thermal profile in the soil (experimental
global radiation at surface level with clear values).
sky. 0, soil surface temperature.
atmospheric downward IR radiation. 0, soil temperature at the lower limit of the
net radiation at ground surface level. integration grid.
IR radiation emitted by soil surface. K thermal diffusivity for soil = 1/C.
dry bulb thermometer temperature at level 2 L thermal conductivity of soil.
m. v cloudiness.
mean value (at daytime scale) of T.,. pa air density at level 2 m.
wet bulb thermometer temperature at level 2 c7 the Stefan constant
m. r diurnal period.
REFERENCES
Assamoi, P. 1984a. Numerical weather forecasting in energy balance, moisture availability and thermal
western Africa: influence of sub-grid scale parametri- inertia. J. Appl. Meteorol. 20. 67-87.
zation on forecasting efficiency. (In French: Carslaw, H. S. and Jaeger, J. C. 1978. Conduction of
Prevision numerique du temps en Afrique de IOuest: heat in solids. Oxford University Press (2nd edition).
influence de la parametrisation de processus sous- Coulibaly, Y. 1981. Local behaviour of heat and mass
maille sur la performance des modeles de prevision.) surface fluxes at several time scales in the tropical
These de Doctorat es-Sciences Physiques n. 78, region (during WAMEX 1979). (In French:
Universite dAbidjan, Ivory Coast. Evolution locale a differentes echelles de temps des
Assamoi, P. 1984b. Numerical investigation of west flux de chaleur et de masse en zone tropicale (periode
African disturbances during monsoon. (In French: WAMEX 1979).) These de Troisieme Cycle no. 663.
Etude numerique des perturbations sur la region L.A.M.P.. Universite de Clermont-Ferrand, France.
Ouest-Africaine en periode de mousson.) Journal Deardorff. J. W. 1978. Efficient prediction of ground
de Recherches Atrnosphtriques 2.8 1-94. surface temperature and moisture with inclusion of a
Blackadar. A. K. 1976. Modeling the nocturnal boun- layer of vegetation. J. Geophys. Res. 83, 1889-1903.
dary layer. Proc. of the Third Symposium of A!- De Vries, D. A. 1966. Thermal properties of soils. In:
rnospheric Turbulence, Diffusion and Air Quality. Physics of plant enuironmenf (ed. W. R. Van Wijk).
Amer. Meteorol. SOC..Boston, 46-49. North Holland, Amsterdam, 21&235.
Budyko, M. 1. 1974. Climate and life. Academic Press. Greenhut, G. K. 1982. Stability dependance of fluxes and
New York. bulk transfer coefficients in a tropical boundary layer.
Carlson, T. N., Dodd. J. K., Benjamin, S. G. and Boundary Layer Meteorol. 24.253-264.
Cooper, J. N. 1981. Satellite estimation of the surface Hervier, R., Soulage, M. and Barthout, J. L. 1979. The
Tellus 3 7 8 (1985), 2
CALCULATION OF GROUND TEMPERATURE AND FLUXES 77
L.A.M.P. automatic micrometeorological station for Publishing Company, Amsterdam, The Netherlands.
surface energy budget measurements. (In French: Reed, R. K. and Halpern, D. 1975. Insolation and net
"Station micrometeorologique automatique du long-wave radiation of the Oregon coast. J. Geophys.
L.A.M.P. pour la mesure du bilan energetique du Res. 80,839-844.
sol".) Technical report no. 24. L.A.M.P., Universite de Rockwood, A. A. and Cox, S . K. 1978. Satellite-inferred
Clermont-Ferrand, France. surface albedo over northwestern Africa. J . Atmos.
Laval, K., Sadourny, R. and Serafini, Y. 1978. Soil Sci. 35, 5 13-522.
surface energy and water budgets formulation in a Saugier, B. and Ripley, E. A. 1978. Evaluation of the
general circulation model. (In French: "Formulation aerodynamic method of determining fluxes over
des bilans energetique et hydrologique a la surface du natural grassland. Q. J. R. Meteorol. Soc. 104,
sol, dans un modele de circulation generale".) Reprint 257-270.
of the National Colloquium on "Mecanismes de Washington, W. M. and Williamson, D. L. 1977. A
Transfert d'energie et de masse entre Sol et At- description of the NCAR global circulation models.
mosphere" (A.S.P. Evolution des climats CNRS- Methods Comput. 17, 1 1 1-1 72.
DGRST) (ed. F. Becker and C. Pastre). 1-40, Zhang, D. and Anthes, R. A. 1982. A high-resolution
Paltridge, G. W. and Platt, L. M. R. 1978. Radiative model of the planetary boundary layer-sensitivity
processes in meteorology and climatology. Deuelop- tests and comparisons with the SESAME-79 data. J.
ments in atmospheric sciences. 5. Elsevier Scientific Appl. Meteorol. 21, 1594- 1609.