used because farmers can choose to use one AI or another. The national sales made
by distributors’ database (BNV-d), which includes pesticide sales, are another
source. This database could be used to spatialise pesticide use [23, 24], but data
have been available only since 2008.
To better define model inputs, a “hybridization approach” using national
resources and local data was developed according to two methods [6]. Since the
Orgeval catchment is only about 100 km
2 , all agricultural practices were monitored
over a period as long as possible for the whole watershed. The Vesle catchment is too
large, and therefore only herbicide use has been investigated within a 30-year period
on targeted practices (weeding vineyards and corn) for which the use of these AIs
was approved. Data were collected at a municipal scale.
3.1 The Orgeval Case
For the Orgeval basin, other data were available to refine the spatial accuracy of the
annual rotation. First, rotation tables were provided directly by farmers, describing
their crop succession on their plots (771 plots accounting for 45% of the landscape
area, dating from 1970 to 2005). Moreover, data from the French Land Parcel
Identification System (LPIS) geographic database [25] has been available
since 2006.
Here, we chose to collect records of their past practices directly from farmers
located in the study area. These documents, called “lowland logbooks” or “field
logs”, help to get closer to farmers’ “real” practices and to reconstruct the dynamics
of phytosanitary practices over 20 years for a representative sample [26]. As an
example, the chart in Fig. 2 shows that atrazine was long the main AI applied on
maize until its ban in 2003. However, the chart still shows some applications beyond
this date, showing the authenticity and low bias of the data recorded. High-resolution
data could capture spatial variability but at the cost of intensive computing
[27, 28]. Therefore, we assumed a type of simplification: to keep confidential and
anonymous data and also to simplify modelling input; these data were averaged by
crop and by year for the whole basin in the final database.
This APOCA (Agricultural Practices of the Orgeval Catchment Area) database
was used to quantify the AIs applied per crop per year since 1990. It also includes a
large number of indicators to monitor changes in phytosanitary pressures [26]. Comparison of these indicators shows that the phytosanitary pressure is not decreasing
overall even if applied quantities per hectare are lower.
3.2 The Vesle Case
For the Vesle study, prescriptions issued by local agricultural development agencies
were used to better indicate actual farmer pesticide use. The technical magazine Le
146
H. Blanchoud et al.
by distributors’ database (BNV-d), which includes pesticide sales, are another
source. This database could be used to spatialise pesticide use [23, 24], but data
have been available only since 2008.
To better define model inputs, a “hybridization approach” using national
resources and local data was developed according to two methods [6]. Since the
Orgeval catchment is only about 100 km
2 , all agricultural practices were monitored
over a period as long as possible for the whole watershed. The Vesle catchment is too
large, and therefore only herbicide use has been investigated within a 30-year period
on targeted practices (weeding vineyards and corn) for which the use of these AIs
was approved. Data were collected at a municipal scale.
3.1 The Orgeval Case
For the Orgeval basin, other data were available to refine the spatial accuracy of the
annual rotation. First, rotation tables were provided directly by farmers, describing
their crop succession on their plots (771 plots accounting for 45% of the landscape
area, dating from 1970 to 2005). Moreover, data from the French Land Parcel
Identification System (LPIS) geographic database [25] has been available
since 2006.
Here, we chose to collect records of their past practices directly from farmers
located in the study area. These documents, called “lowland logbooks” or “field
logs”, help to get closer to farmers’ “real” practices and to reconstruct the dynamics
of phytosanitary practices over 20 years for a representative sample [26]. As an
example, the chart in Fig. 2 shows that atrazine was long the main AI applied on
maize until its ban in 2003. However, the chart still shows some applications beyond
this date, showing the authenticity and low bias of the data recorded. High-resolution
data could capture spatial variability but at the cost of intensive computing
[27, 28]. Therefore, we assumed a type of simplification: to keep confidential and
anonymous data and also to simplify modelling input; these data were averaged by
crop and by year for the whole basin in the final database.
This APOCA (Agricultural Practices of the Orgeval Catchment Area) database
was used to quantify the AIs applied per crop per year since 1990. It also includes a
large number of indicators to monitor changes in phytosanitary pressures [26]. Comparison of these indicators shows that the phytosanitary pressure is not decreasing
overall even if applied quantities per hectare are lower.
3.2 The Vesle Case
For the Vesle study, prescriptions issued by local agricultural development agencies
were used to better indicate actual farmer pesticide use. The technical magazine Le
146
H. Blanchoud et al.