Page 93 - Modelisation du devenir des pesticides...

3.3 Results and discussion

present large variations with no clear seasonal signal. As in the model, the water
is drained above the field capacity ; the simulation presents fewer variations than
the observation especially during rain events. Overall, the simulation of the water
profile and drainage flux shows some discrepancies compared to the observations,
although the drainage flow dynamic is caught by the model. This will certainly
impact the solute transfer, either conservative or reactive. As capacity model have
simplified water transfer scheme, more complex processes such as preferential flow
ore capillary rise are not taken into account. The field capacity is the most influen-
tial parameter for water transfer and must be defined with accuracy. 67 Moreover
the soil parameters measured by Boesten and Van der Pas 55 were only extracted
from one plot in the field at the beginning of the monitoring and did not take the
spatial variability at the field scale into consideration. The discrepencies between
the predictions of STICS and water drainage measurements at Kerlavic are in the
same range as previous results obtained with the capacity model Pelmo at Tor
Mancia 26 or with the Richards type models LEACHP and WAVE at Weiherbach.
33,34

3.3.3 Bromide content

As presented in section 2.3, the bromide was used at the three sites as a conserva-
tive tracer. The observed and simulated bromide profiles or fluxes were compared
to assess the solute transport. At Vredepeel, a solution of Bromide was applied at
the end of November 1990 using a pesticide sprayer. bromide distribution profiles
at three sampling dates are presented in Fig. 3.3.2. The simulated bromide profiles
are more in agreement with the observations than the soil water profiles (Table
3.6) : over the full 6 sampling dates, there is almost no bias, the correlation reaches
0.65 and the efficiency is 0.43. The model represents the bromide dynamics : a peak
of bromide concentration at the top of the soil 22 days after applications (DAA)
and a flatter peak around 80cm deep at 214 DAA. The discrepancy was larger at
278 DAA (August 27), where the observations show a flat bromide profile, even in
the top 40 cm, while the simulated profile is similar to that of 214DAA. This is
certainly linked to the error in the soil water profile (section 3.3.1).
At Kerlavic, the observed bromide leaching fluxes vary considerably between
replicates for the 2 years. The first year bromide is detected in the leachate of the
lysimeter 4 few days after application, while no bromide is detected for the other
two replicates. The bromide is not detected in the leachate during the earlier time
after application on year 2003/2004. Such discrepancies are commonly observed
in pesticide fate studies performed with lysimeters and are attributed to spatial
variability of soil hydraulic properties. 77-79 But, in Kerlavic, as a restructured soil
was used, such behaviour may also indicate the occurrence of preferential flow.
Consistently with the underestimation of the drainage by the model, the simula-

81