Page 97 - Modelisation du devenir des pesticides...

3.3 Results and discussion

ron leaching is about 3 times greater in the 1st year than in the 2nd. Although
the model strongly underestimates the isoproturon leaching (from about -60% to
-90%), it is able to capture the dynamics between years. The higher flux in the
1st year is linked to two major precipitation events leading to large drainage. The
STICS predictions for isoproturon are not within the spread of the observed data in
2002/2003, whereas the predicted isoproturon leaching is close to the lowest value
of the observed data in 2003/2004. In addition, the model reproduces the isopro-
turon leaching dynamics satisfactorily lysimeter 8 is excluded in year 2002/2003
in consistence with Table 3.6.
Atrazine is assumed to be more strongly adsorbed onto soil particles than iso-
proturon. At the Thiverval-Grignon site, the dissipation of atrazine was previously
simulated by a former version of STICS integrating a pesticide fate module. 20 The
results we obtained are fairly similar : the simulated data match the measured ones
with good accuracy (efficiency ranging between 0.2 and 0.8,Table 3.6) with almost
no bias in the second year and an underestimation of12% to 25% the 1st year.
Overall, the pesticide simulated fates show different patterns depending on the
sites : quite good results (efficiency above 0.2) in Thiverval-Grignon for both years
although this was not the case for the bromide, reasonable results in Vredepeel
(efficiency of 0.22, overestimation by 44%) and poor in Kerlavic (negative efficiency,
underestimation by more than 60%). These results seem to be more connected to
the ability of the model to reproduce the water flow than to the pesticide properties.

3.3.5 Pesticide mass recovery budget

The model gives a similar representation of the processes and the partition bet-
ween degradation, leaching, and the two sorption processes for the 3 sites whereas
this is not possible based on the observations. Fig. 3.3.4 presents the simulated
mass balance of pesticide for the 3 sites over a similar duration and at two inte-
grated depth, of 30cm and one meter.
Table 3.7 compares the mass balance of the 3 sites over the first meter a similar
time after application of 250 days. The recovery mass (sum of liquid and adsorbed
pesticide at equilibrium masses) can be compared to the observations when avai-
lable. The mass balance for bentazone at Vredepeel is clearly different from the
other two. A small fraction of applied pesticide is kept in the soil after a 15 months
period (1%), while a large part is leached (25%). For the top soil, the liquid phase
is dominant at Vredepeel during the first 30 days after application (47%) while it
is much lower in the other two sites (less than 8%). The large amount of Benta-
zone in the liquid phase in Vredepeel is linked to two processes i) weaker linear
adsorption and ii) a weaker degradation (Table 3.4). Although the degradation is
the dominant process on the three sites over a long period (Table 3.7), it has a
slower dynamic at Vredepeel, only 7% of the applied pesticide being degraded on

85