**2.2.4 Integration and aggregation of SCALES data**

SCALES is a large-scale assessment model intended for mapping soil erosion hazard at the finest level of organization of the agricultural area. It supposes an integration of input data at parcel unit scale. Their area can exceed a few hectares but are generally lower than this unit of reference. This scale seems to be the better scale to determine the spatial context of soil erosion in Basse-Normandie. The layer of parcel units exists in the form of a vectorized and geo-referred database and such data came from the Inventory of Common Agricultural Politics given by the agricultural administration. Data inform on the land cover types in 2006 (source: Rpg\_anonyme\_014\_AUP\_2006). The parcels are classified in the three following: Grassland and arboriculture parcels (26,500 parcels, 111,000 ha) Temporary grassland parcels (13,200 parcels, 99,600 ha) and Crop parcels (25,800 parcels, 154,400 ha).

The integration of data in the parcel units requires beforehand a mapping of input data under raster format. We used the principle of allocating for each parcel unit and for each input data type, only one mode. When parcel unit holds several modes, we chose using a decision rule, to select the spatially most dominant mode. The application of the dominance rule has been carried out using the module *Spatial Analyst* in ESRI ® ArcView Gis 9.2. Obtained maps concerning the rainfall erosivity, the potential sensibility of areas and the soil erosion hazard are vector maps. Those are transformed to raster format for incorporating the output data in larger vector units as administrative units (approximately 800 ha) or as hydrologic units (approximately 2,000 ha).

SCALES: An Original Model to Diagnose Soil Erosion Hazard

agricultural input data of the SCALES model.

data.

arboriculture areas.

**2.3.4 Soils** 

and Assess the Impact of Climate Change on Its Evolution 235

Adapted statistical treatments allowed us to draw up a typology of dairy farm (9 types) and non-dairy farm (13 types). A statistic aggregation (by summation) has been realized in order to assign them to the small administrative units (municipality scale). That led to the characterization of repartition profiles of different farm types for each municipality of Calvados which counts 706 of them. The following stage consisted in operating an Ascending Hierarchical Cluster followed by K-means method in order to reach a typology of farms (12 types) according to type repartition profiles (Bermond, 2004). Each type refers to modes of farm management, and to specific agricultural practices. A local farmer practice survey has been carried out in this direction, which enable us to produce our own

After selecting a sample of municipalities for each farm type, we interviewed the farmers about soil work methods used, plot localization and farm characteristics. This investigation showed that types of farm had notably evolved between 2000 and 2007. Thus, interviews have been used to update the 12 types of farm and to specify the current agricultural practices. This procedure has been applied to the sampled administrative units and, by extrapolation, generalized with all municipalities. Knowledge of crop rotations and management of intercrops allowed us to determine various modes concerning these

For crop rotations, the types are: winter crops, dominance of winter crop, balance winter crops/ spring crops, dominance of spring crops, spring crops. The passage of the first to the last type represents insofar the lengthening of the period during which soil is not protected. The duration of this period is in this way lower than 4 months for rotations based on winter

Regarding the management of intercrops numerous publications underline the influence of different practices on soil erosion risks (Auzet, 1987; Martin, 1997; Martin 1999; Baumhardt and Jones, 2002; Le Bissonnais et al*.*, 2002; Lipiec et al., 2006; Strudley et al., 2008). Therefore the creation of a temporal plant cover like oilseedrape or mustard, in the period between two crops, will effectively protect the soil against run-off erosion. This measure will be less effective in case the crop partially covers the soil like for example rye (concept of scarce plant cover). One also considers that the wheat stubble correspond to this concept. With the absence of a temporal crop, soil tillage will permit to temporally reduce the erosion risk because of a better infiltration and a higher soil roughness. More the tillage operations are deeper, more effective are infiltration and soil roughness against soil erosion. The most unfavorable condition occurs when there is no tillage during the intercrop period remaining the soil bare. These different practices between crops or their absence (plant cover, scarce plant cover, deep ploughing, superficial ploughing, bare soil) are respectively comparable with the levels of pressure 1, 2, 3, 4, 5. The level 0 corresponds at the grassland and

To achieve the aim of a fine diagnosis of soil erosion hazard, it was necessary to have a sufficiently precise soil database. It was not conceivable to exploit the Soil Geographical Database of France at scale 1 : 1 000 000. The regional BDSol-250 on a 1 : 250 000 scale

(source INRA) does not exist. So we decided to create our own data.

crops and reaches a duration of 7 months in case of a succession of spring crops. Those are respectively comparable with the levels of pressure 1, 2, 3, 4, 5.
