Page 107 - Modelisation du devenir des pesticides...

3.5 Conclusion du chapitre

!! "#$%&! "'%(! &##! $)*+,-./! $)*+,%01! %2,-./! %2,%01!
#)34! 567487! 94:5! ;;48<! ;=466! =74=7!
2>,=9,10+! 56?468! 94:6! @;7477! 5=4<<! ;946=!
2>,=9,1*A! 56648;! 94:?! ?4;9! 6647;! =84<7!
2>,=9,1)*+! 56:4;5! 94:7! @84?;! =:457! :?4?<! ?:45<! ==495!
BCD,10+! 578455! @94=8! =646<! ;?49:! ?:469!
BCD,1*A! <9=47:! @564?5! 5694;:! 8:45<! :;6469!
BCD,1)*+! ;:;4?8! @<4:?! 754==! =747?! 58547=!
EC+/F*+F!-$! 556479! 94;?! ;:46:! ;;4=8! ;?4<=!
(GHI*JK0D!L*I*1! 5;54:=! 9457! 5:47<! 6=468! ==49?!

Table 3.11 – statistical score obtained at Vredepeel for the sensistivity test ;
NRMSE (%) ;err (%) ; Mean (µg/dm ) ; SD Standard Deviation
3
(µg/dm )
3

tested in Tiktak 69 and Vanclooster 71 , for both the calibrated and uncalibrated
run, for 3 over the 6 available profiles. Bentazone profile estimated by Stics are
relatively close to simulations performed with PLM 28 , which might be linked to
the use of the same value for field capacity and wilting point. However, there are
some deviations between the two models which may partly be attribute to the
different adsorption and degradation parameters 28 . As already stated, the soil
transfer in Stics is based on a capacity models (drainage occurs as soon the filed
capacity is reached). Compared to other capacity models tested at Vredepeel, the
bentazone residues predicted by Stics are better reproduce (i.e., less overestimated

64,65 ). The statistical score obtained by Stics are in the range of the other ones,
with the exception of the efficiency on the bentazone profiles that is in the range
of uncalibrated values, but not in the range of the calibrated value, which is also
consistent with the fact that the reference run with stics was not calibrated.

3.5 Conclusion du chapitre

Les résultats et les discussions présentés dans ce chapitre ont mis en évidence
la difficulté associée à l’évaluation du modèle sur différents sites et molécules aux
caractéristiques contrastées.
Par ailleurs, la complexité de l’évaluation est renforcée par le caractère lacu-
naire des jeux de données et la variété des méthodes d’acquisition utilisées sur les
différents sites d’études.
Ce travail souligne également le manque de données sur les sites expérimentaux

95