Description of the selected region: the Sahel and the site of HAPEX-Sahel experiment
The region analysed in this study is located in
the Sahel (Africa). The Sahel occupies a narrow zone, forming a strip about
400-600 km wide, bounded on the North by the Sahara and on the South by
the Sudanian vegetation zone (Figure 1). It stretches nearly 6.000 Km across
the entire African continent.
The Sahel climate is characterised by a single,
short, annual rainy season associated with the Northward movement of the
Inter Tropical Convergence Zone (ITCZ).
Despite of this severe climate, the Sahel is an
area extensively exploited by the humans; thus, much of the landscape of
the Southern and more favourable mid-Sahel consist of fields and extensive
areas of fallow bushland in various stages of regrowth.
Figure
1. The selected region: the Sahel in Niger near Niamey (the capital city).
An area belonging to the Nigerian Sahel, of 1° ´ 1° (latitude 2°-3° East and longitude 13° - 14° North, corresponding to a surface of almost 100 ´ 100 Km2) was selected in 1990-1992 for the HAPEX-Sahel field experiment. All the three sites are located in this area. Niamey, capital city of Niger, is in this area too. The coordinates of the sites are: Fallow 13.244 °N 2.244 °E, Millet 13.241 °N 2.999 °E, Tiger 13.198 °N 2.239 °E.
The LSPM initialisation
The input data required by LSPM have been taken from the HAPEX-Sahel field experiment. As already mentioned, LSPM needs the following data: air temperature, atmospheric pressure, precipitation rate, solar radiation or cloudiness, specific or relative humidity and horizontal wind components. For all simulations, we have used 8 soil layers whose thickness was respectively 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 1.0, 1.5 m. The entire dataset of the HAPEX-Sahel experiment consists of the observations referring to the period from middle of 1990 to late1992. From this dataset we selected the data (including the input) of the period from 17 August 1992 to 6 October 1992 (the rainy season) during which the soil moisture was measured in the three sites.
The results
Among all outputs calculated by LSPM, we selected those allowing to make the comparison with the observed data of the following variables: soil moisture at different depths, net radiation, sensible and latent heat fluxes. The following Subsections give a detailed account of the main results achieved in the three selected sites.
The Fallow site
Figure
2 Observed soil moisture in Fallow site, at 10 cm, 20 cm, 120 cm and 180
cm below the surface, respectively.
From Figure 2, which shows the observed soil moisture in Fallow site, it appears that, after the heavy rain period, the soil moisture responds very quickly to the precipitation even at the deepest two levels. For instance, on August 22 and 31 (see also Figure 3), the daily averaged precipitation rate reached 37 mm/day and 49 mm/day, respectively. Obviously, the soil moistures at 10 cm and 20 cm got wet immediately, while the soil moisture at 120 cm took only 1 day to feel the surface water maximum and that at 180 cm took about 2 or 3 days. This fact proves that the soil in this region has very high hydraulic conductivity. According to our experiences, it is quite unlikely that sandy soil parameters of CH78 can bring about such a high conductivity. In fact, LSPM failed when sandy soil parameters of CH78 were used in LSPM to simulate the soil moisture profile, as shown in figures 4(a,b,c,d) referring to LSPM simulation of soil moisture with the sandy soil parameters of CH78.
Figure
3 Daily averaged precipitation rate (mm/day) in Fallow site
 |
Figure 4(a,b,c and d) Observed versus simulated soil moisture in Fallow site, at 10, 20, 120 and 180 cm below the surface, with soil parameters of CH78 |
From figures 4(a and b), referring to the first
two layers, we can see that the simulation of the soil moistures with soil
parameters of CH78 greatly overestimates the observation by more than 0.05
m3m-3 . The differences increase when soil becomes drier. Of course, simulations
and observations are in phase since these soil layers are near the surface.
However, when we look at figures 4(c and d), not only the simulations overestimate
the observations, but they are also out of phase. In Figure 4(c),
the soil 120 cm below the surface took almost 10 days to respond the rainfall
instead of 1 day, as shown by the observation (Figure 2). Because of the
failure of such a simulation, we tried to use for the simulation new sandy
soil parameters measured in Grugliasco (Turin, Italy), where a hill rich
of pure sandy land coming from an ancient glacial deposit is present. The
Grugliasco sand is characterised by an extremely high hydraulic conductivity.
We call it “pure sand”. Table II shows the new soil parameters, against
the corresponding ones of CH78.
|
|
|
|
CH78 |
|
|
|
|
|
|
|
|
|
|
According to the parameterization of hydraulic conductivity,
under moderate soil moisture condition, the hydraulic conductivity of the
pure sand is about 5 times larger than that of CH78, and under dry soil
moisture condition, the hydraulic conductivity of the pure sand is about
100 times larger than that of CH78. By using these sandy soil parameters
of the pure sand, we simulated the soil moisture again in Fallow site.
Figures 5(a,b,c and d) show the results.
 |
Figure 5(a, b, c and d) Observed versus simulated soil moisture in Fallow site, at 10, 20, 120 and 180 cm below the surface with observed “pure sand” soil parameters in Grugliasco, Italy, without considering the contribution of underground water vapour to the surface water budget |
In this case, the soil moistures look well simulated.
The simulation values not only show a good agreement with the observations,
but they are also in phase at all levels underground. A noticeable feature
in Figure 5(a) is that the simulated soil moisture (at 10 cm under the
surface) shows much higher values during the short raining period. This
is not necessarily a fault of simulation. On the contrary, this is because
the observation of soil moisture was made after rain stopped. Due to the
very high hydraulic conductivity, the soil moisture near the surface decreases
very quickly. A lot of water near the surface has infiltrated into deeper
levels before the observation was made.
Even if the LSPM simulation shows a sharp improvement
when the observed soil parameters of the pure sand are used, there are
still some problems in Figures 5(a and b). From Figure 5(a), we can
find that, at the end of simulation, the first layer of soil moisture almost
goes to zero rather than to the observed value of 0.02 m3m-3.
When the first layer of soil moisture q1 becomes very dry, surface resistance
that accounts for water vapour transfer between the soil and surface (which
in turn is a function of q1) becomes very high. This reduces the surface
evaporation over bare soil surface and hampers the deep soil moisture from
going up to the surface. In this case, only the transpiration is left.
In fact, from Figure 5(b), we can see that at the end of simulation the
soil moisture at 20 cm below the surface slightly overestimated the observation.
These problems are due to the fact that the model was not able to extract
water from dry soil, after evaporation of most of surface liquid water,
because the scheme of underground water vapour was not activated.
When this scheme is activated, we got Figures 6(a, b, c and d).
 |
Figure 6(a, b, c and d) Observed versus simulated soil moisture in Fallow site, at 10, 20, 120 and 180 cm below the surface, with observed “pure sand” soil parameters in Grugliasco, Italy and with considering the contribution of underground water vapour to the surface water budget |
By comparing Figures 6(a and b) with Figures 5(a
and b), we can clearly appraise the improvement of the simulation after
activation of the underground water vapour scheme.
With the observed “pure sandy soil” parameters,
not only soil moistures, but also soil temperature, surface heat fluxes
and net radiation are well simulated. Figure 7 (a, b, c and d) show observed
versus simulated soil temperature at 10 cm under the surface, and surface
sensible heat flux, latent heat flux and net radiation respectively.
 |
Figure 7 (a) Observed versus simulated soil temperature at 10 cm under surface. (b) Observed versus simulated sensible heat flux. (c) Observed versus simulated latent heat flux. (d) Observed versus simulated net radiation. |
Generally speaking, soil temperature is well simulated.
Only during some short period, its simulation underestimates the observation.
As far as sensible heat flux, latent heat flux and net radiation flux are
concerned, continuous observation could hardly be available in Fallow site,
because the instruments used by HAPEX-Sahel exhibited some problems to
work properly when it rained. From the few days of observations, we can
see that the observed sensible heat flux is quite small. Generally speaking,
the daily averaged sensible heat flux is about 30 Wm-2. The observed latent
flux is 3 times larger than the observed sensible heat flux, which demonstrates
the high evaporation in this region. Only after the raining season, when
the soil becomes dry, the observed sensible heat flux starts increasing
and the observed latent heat flux decreasing. They become comparable
at the end of the considered observation period.
LSPM proves to be able to reproduce this phenomenon
very well. The simulations meet the observations properly, when these last
are available. Figure 7c, more than the other three figures, seems to suggest
that the simulated latent heat flux exceeds some time the observed one.
Once again, this is only an apparent disagreement, due to the fact that
the observed values are not as regular and continuous in time as simulation
is.
The Millet and Tiger sites
With the same soil parameters used for Fallow site
and the scheme of underground water vapour set on, LSPM was also applied
to Millet and Tiger sites. Figures 8 (a,b,c and d) show the simulation
of soil moisture in Millet site. Simulation of other physical quantities
were not carried out here, because of shortage and randomness of relevant
observations.
 |
Figure 8 (a, b, c and d) Observed versus simulated soil moisture in Millet site, at 10, 20, 120 and 180 cm below the surface, with observed “pure sand” soil parameters in Grugliasco, Italy and with considering the contribution of underground water vapour to the surface water budget |
Generally speaking, the simulation in Millet site
shows good agreement with the observation at all levels. The careful readers
might find that the simulation at the deepest levels [Figures 8 (c and
d)] did not follow the observation as precisely as it did in Fallow site
[Figures 6 (c and d)]. Nevertheless, this result still keeps good enough
for practical applications.
The simulation of soil moisture in Tiger site,
on the contrary, shows a different feature. Figure 9 (a, b, c and
d) Observed versus simulated soil moisture in Tiger site, at 10, 20, 120
and 180 cm below the Surface, with observed “pure sand” soil parameters
in Grugliasco, Italy and considering the contribution of underground
water vapour to the surface water budget
 |
Figure 9 |
The agreement of simulation with observation at
the levels near the surface is not so bad [Figures 9 (a and b)], except
for the first three days. But the simulation at the deepest levels [Figure
9 (c and d)] is unacceptable. What we could not understand is why the observation
values at deep levels in Tiger site keep very high after the rain stopped.
They almost reached and held the saturated value. Probably, the water
table level here is quite high. If this is true, the observation would
refer to a different phenomenology from the one simulated by our model.
Conclusions
According to both the observation and model simulation
of soil moisture in Sahel (Niger), it proves that the hydraulic conductivity
in the studied sites is higher than one prescribed by CH78 for sand soil.
Because of the high hydraulic conductivity, the surface soil moisture becomes
very dry shortly after rain stops. Under this circumstance, it is necessary
to take account of the underground water vapour contribution to the surface
water budget. Our Land Surface Process Model (LSPM) has been upgraded with
this scheme to simulation soil moisture under this extreme condition. The
simulation results show great improvement with this scheme.
