MODELING THE WATER BALANCE COMPONENTS OF THE SOYBEAN CANOPY BY SOIL-VEGETATIONATMOSPHERE TRANSFER MODEL - Soybean and Wheat Crops: Growth, Fertilization, and Yield - page 57

between the simulation and the observations. Moreover, the LAPS correctly describes the
trend of the soil moisture in a qualitative sense. The curve representing simulated values of
the total soil water content over 1.6 m depth is very close to the observations. At the
beginning of the growing season (early May), the soil was loosing the water intensively since
the evapotranspiration was the dominant process. The good agreement between the simulated
and observed values is extended up the end of the growing season.
It has been increasingly apparent that the water fluxes from a natural surface (covered by
vegetation or a bare soil) simulated by the land surface scheme and their correct partitioning
into horizontal and vertical water fluxes, and evaporation, are sensitive to the procedure for
calculating its temperature [53]. Specification of the deep soil temperature cnnot be easy done
when a land surface scheme provides the lower boundary condition in atmospheric models
covering the domain of various soil textures. A wrong specification of deep soil temperature
introduces an error in calculating the ground temperature and evaporation [24] and finally the
incorrect partitioning of the surface water into water balance components. These errors are
even more pronounced during long term integration. Mihailovic et al. (1998) [52] have
discussed how the changes of the deep soil temperature in the restore term of the “force-
restore” equation affect the partitioning of: (i) the surface energy into the sensible and latent
heat fluxes; and (ii) the land surface water into water balance components. Basically, there are
several possibilities for determining the deep soil temperature. For instance, it can be (i)
specified as a constant, (ii) calculated from a prognostic equation or (iii) calculated as a
running mean of ground temperature from the previous day as it is done in the LAPS,.
Integrated water balance components (cumulative values) and soil water change for the first
120 days of integration at Caumont (France) are given in Figure 7. Calculated, cumulative
values are: 151.8 mm for the evaporation, 239.5 mm for the horizontal and vertical flows
through the corresponding borders and 22.3 mm for the soil water. Comparison these values
with the observed ones (in the brackets of this figure) shows that the LAPS correctly simulate
water balance components.
3.3. Simulation of Water Balance Components
Model outputs of latent heat flux (evaporation from bare soil fraction and water
intercepted by leaves and transpiration) for Day of Year (DOY) 150-155 were compared with
measurements in a soybean field at Coumont (France) using the aforementioned HAPEX data
set. The parameters used in this simulation are given in Mihailovic et al. (1998) [52]. Monthly
mean leaf area index, LAI , the fractional vegetation cover, σ f , overall surface roughness
length, z 0 , the canopy roughness length z 0c , the zero-displacement height, d , and the canopy
height, H , are as specified in the Mihailovic et al. (1998) [52]. These quantities are
interpolated to smaller time intervals when required. The roots of soybean plants were
assumed to be shallow and are distributed mainly in the top 0.5 m. The year was divided into
the bare soil period (January-April, October-December), the transition period (May) and the
growing season (June-September). For the bare soil period, there were no roots; for the
transition period, it was assumed that the top 0.1 m soil layer contains 70% of the roots and
the soil layer between 0.1-0.5 m contains the rest i.e. 30%; for the growing season, 60%, 30%
and 10% roots were assumed to be in the soil layers 0-0.1, 0.1-0.5 and 0.5-1.6 m,
respectively. Figure 8 depicts comparison between the calculated diurnal variations of latent