The next step for evaluating the contamination of surface and groundwaters at the
Seine River basin scale will be based on mechanistic modelling, as already
implemented on the Vesle basin.
7 Conclusion
Depending on the size of the watershed under study, pesticide use and database
monitoring differ.
• At the plot scale, crop rotations and phytosanitary use and application conditions
are known and can be combined for precise and adapted monitoring.
• In a catchment area such as Orgeval (100 km
2 ), the agricultural landscape is more
complex. The observed contamination corresponds to major pesticide use. However, homogeneity is still observed due to a similar pedoclimatic context. The
response time of the soil-groundwater-river system is fast, and contamination
appears quickly and abruptly at the outlet because of drains. Major pesticide
inputs can be identified using farmers’ logbooks.
• In the Vesle basin (1,000 km
2 ), the water transfer time from groundwater to the
river is longer. The landscape is contrasted, resulting in heterogeneous crops in
the basin (alfalfa, annual crops and vineyards). Spatialisation of uses will be of
particular importance in this context, and we need to combine agricultural
practices (e.g. pesticide treatment) and soil characteristics. In this case, the
random distribution of crops considered in our modelling approach of basins
may have a considerable impact on the model’s outputs. Vineyards receive
specific treatments and are well identified in the Corine land cover database; its
pesticide applications were therefore correctly located. Modelling also required
estimating inputs over at least 30 years, to include the transfer time to the outlet.
This retrospective raised the question of the initialisation of the model, which
takes a considerable part in the uncertainty of the simulation results. Nevertheless,
focusing on specific weeding practices in vineyards and field crops allowed us to
estimate the uses of atrazine and to model its transfer. However, working at the
sub-annual level would not be realistic at this time.
• At the scale of the Seine basin (75,000 km
2 ), the hydrogeological context is
complex, with several superimposed aquifers, different response times and a
patchwork of cropping systems. Monitoring databases are those assembled by
the national authority for European framework directive monitoring [15, 81], and
they allow only scarce and discontinuous monitoring of pesticide contamination.
However, considering only the residues of atrazine, which has been widely used
throughout the basin and frequently detected, it is possible to show that some
aquifers still have an increasing contamination of atrazine residues despite its
prohibition in 2003. Particularly vulnerable groundwaters such as the Champigny
aquifer still show an increase in DEA levels, and other studies have shown
156
H. Blanchoud et al.
Seine River basin scale will be based on mechanistic modelling, as already
implemented on the Vesle basin.
7 Conclusion
Depending on the size of the watershed under study, pesticide use and database
monitoring differ.
• At the plot scale, crop rotations and phytosanitary use and application conditions
are known and can be combined for precise and adapted monitoring.
• In a catchment area such as Orgeval (100 km
2 ), the agricultural landscape is more
complex. The observed contamination corresponds to major pesticide use. However, homogeneity is still observed due to a similar pedoclimatic context. The
response time of the soil-groundwater-river system is fast, and contamination
appears quickly and abruptly at the outlet because of drains. Major pesticide
inputs can be identified using farmers’ logbooks.
• In the Vesle basin (1,000 km
2 ), the water transfer time from groundwater to the
river is longer. The landscape is contrasted, resulting in heterogeneous crops in
the basin (alfalfa, annual crops and vineyards). Spatialisation of uses will be of
particular importance in this context, and we need to combine agricultural
practices (e.g. pesticide treatment) and soil characteristics. In this case, the
random distribution of crops considered in our modelling approach of basins
may have a considerable impact on the model’s outputs. Vineyards receive
specific treatments and are well identified in the Corine land cover database; its
pesticide applications were therefore correctly located. Modelling also required
estimating inputs over at least 30 years, to include the transfer time to the outlet.
This retrospective raised the question of the initialisation of the model, which
takes a considerable part in the uncertainty of the simulation results. Nevertheless,
focusing on specific weeding practices in vineyards and field crops allowed us to
estimate the uses of atrazine and to model its transfer. However, working at the
sub-annual level would not be realistic at this time.
• At the scale of the Seine basin (75,000 km
2 ), the hydrogeological context is
complex, with several superimposed aquifers, different response times and a
patchwork of cropping systems. Monitoring databases are those assembled by
the national authority for European framework directive monitoring [15, 81], and
they allow only scarce and discontinuous monitoring of pesticide contamination.
However, considering only the residues of atrazine, which has been widely used
throughout the basin and frequently detected, it is possible to show that some
aquifers still have an increasing contamination of atrazine residues despite its
prohibition in 2003. Particularly vulnerable groundwaters such as the Champigny
aquifer still show an increase in DEA levels, and other studies have shown
156
H. Blanchoud et al.