Berien Elbersen Maarten Hilferink Wim Nieuwenhuizen Marta P

1
allocation of ELPEN farm systems

• Berien Elbersen Maarten HilferinkWim
  Nieuwenhuizen Marta Pérez-Soba Janneke Roos
• Alterra, The Netherlands
• Udine, 22 nov 2002

2
To determine environmental and socio-economic
impacts we need
real farm data grouped in farm systems (survey
information from representative farms to
understand and model the farming practices and
their behaviour) to locate the farm systems (to
connect them to the environment they influence,
e.g. social and (a)biotic circumstances,
landscape, biodiversity) to determine how farm
practices have impact on the environment (employm
ent, eutrophication, acidification, habitat
maintenance etc.)
3
Allocation exercises at a national level

• Objective validation and improving the methods
  for allocation of animal categories
• Using collecting statistical data on a lower
  regional level than NUTS 2, e.g. canton in
  France, grids of 5 km² in Denmark
• Countries Denmark, France, The Netherlands,
  Portugal, Spain, UK (England Wales and Scotland)

4
Method
1) Regression analyses to estimate the influence
(weight) of different variables on the number
of animals 2) Use the regression formula to
improve the allocation approach 5 in
ELPEN 3) Compare these results with national
data on the number of livestock on a lower
territorial level 4) Compare the results of the
old approach 5 with the national data and
find out whether the new approach has better
results.
5
Spatial variables
Animal categories (nr per sub-region)
CORINE Land cover classesClimateAltitudeSoil
typeAccessibility (Denmark)
Dairy cowsSuckler cowsSheep dairySheep
meatGoats
6
Results France
Dairy Cows France 1 Pastures2 Complex
cultivation patterns3 Discontinuous urban
fabric4 Broad-leafed forest Dairy Sheep
France 1 Natural grassland2 Complex cultivation
patterns3 Transitional woodland/shrub4
Non-irrigated arable land5 Pastures Meat Sheep
France 1 Pastures2 Land principally occupied by
agric.3 Broad-leafed forest4 Sclerophyllous
vegetation
Cantons in France
7
Results Denmark
Dairy Denmark 1 Non-irrigated arable land2
Complex cultivation patterns3 Pastures4 Land
principally occupied by agric. Sheep Denmark 1
Pastures2 Land principally occupied by agric.3
Complex cultivation patterns4 Non-irrigated
arable land Conclusions of comparison Approach
5 50 of grid cells correct Using results regr
analysis 55 to 60
5km grid data
8
Conclusion for the allocation in ELPEN
The allocation should be improved The allocation
rules must be spatially and farm-/livestocktype
specific Use of national data is not
practical So Use all the specific spatial and
farm/livestock data we have in the ELPEN system
for the allocation
9
How to allocate (groups of) ELPEN farm systems on
grid cell level?

• The available statistical data on the ELPEN farm
  systems are FADN data on land use and livestock
  on HARM1 or ELPEN region level
• But also LFA altitude level relation with EU
  farm type
• So
• Use FSS data on corresponding EU farm types to
  desaggregate to HARM2 level
• Use Corine/Pelcom land cover data in
  correspondence with the land use data to
  desaggregate to grid cell level
• Use LFA Altitude geodata on grid cell level to
  allocate the in FADN provided of LU, UAA
  within these regions

10
Why do we need ELPEN regions?

• ELPEN region 7
• 17 farm system types
• 512896 Livestock units
• 456098 ha of UAA

• Total of HARM regions
• 11 farm systems (6 are missing)
• 56 of LU in ELPEN region
• 49 of UAA in ELPEN region

11
Data
Allocation items
Aggregation level
12
Allocation of ELPEN farm system (group)s (EFS)1)
Desaggregate to HARM2 level using FSS data
Efs D in HARM1-y
13
Allocation of ELPEN farm system (group)s (EFS)2)
Desaggregate to GRID level using Corine/Pelcom
Efs D in HARM1-y
Corine/Pelcom land cover groupsin HARM1-y
14
Allocation of ELPEN farm system (group)s (EFS)3)
use LFA Altitude geodata to complete the
allocation
Efs D in HARM1-y
15
(No Transcript)