Page 103 - Modelisation du devenir des pesticides...

3.4 Conclusion

were available. Therefore, different experimental approaches are now used in the
experimental basin of the Orgeval, 86 completed by the design of an agricultural
practices database oriented towards pesticide application. 87 All these data will
be used to apply the STICS model, and also the study of Flipo et al., 88 who
focused only on nitrogen pollution in the hydrosystem, in order to treat the two
main sources of agronomic diffuse pollution, nitrogen and pesticides.

ACKNOWLEDGMENTS
This study was supported by the PIREN SEINE and the EC2CO Phyt’Oracle project fundedby
the CNRS. The Naturalia and Biologia association and Alterra are thanked for supporting the
collaboration between UPMC and WUR universities. The authors also wish to thank Frederik
Van Den Berg and Jos Boesten for supplying the Vredepeel field-data and for their advices on
the model application. The authors would like to thanks the AgroImpact team of INRA for their
advices and explanations, and more especially Elsa Coucheney, Nicolas Beaudoin and Bruno
Mary for providing field data from Kerlavic and guidances for the crop model parametrization.

Appendix A : sensitivity test

The aims of tthe pesticide trasnfer module (PeStics) included in STICS is to be
applied at the basin scale in order to estimate the diffuse pollution associated to
agriculture, either linked to the N sources 5,9-11 and pesticides that is transferred
to the hydrosystem. The assessment presented above on 3 different sites and 3
different pesticides allows getting an idea of the model forces (generecity) and
weaknesses (mostly associated to soil water transfer in complex sites with shallow
aquifer or preferential flow). However, these results were obtained mostly using in-
situ measured parameters or parameters extracted from literature review from the
experimental site. Such in-situ data won’t be available at the catchment scale. In
order to get an idea of the error that could be associated to a poor determination
of the parameters, a sensitivity test is performed. The targeted output used to
estimate the sensitivity is the cumulative areic mass lost via leaching, which is the
one that can reach the hydrosystem. A one-at-a-time sensitivity analysis similar
to Dubus et al., 82 is performed, by varying one input parameter, the others being
kept to their reference values. The sensitivity of two pesticide parameters (KOC
and DT50) is tested by using the minimum, maximum and mean values provided
by Footprint 42 (Table 3.9).
As already stated, the DT50 used in Vredepeel was determined based on labo-
ratory experiment, and is about 3 times larger than the maximum values from
Footprint, while for the 2 other sites, the reference value is close to the mean. Koc
is close to the minimum value for Isoproturon in Kerlavic, and smaller than the
minimum value for the two other sites. Three additional tests were performed in
order to test the impact of non-equilibrium sorption for isoproturon and atrazine,

91