WP 8 CAPRI GIS Link

1
WP 8 CAPRI GIS Link
Agricultural Policy The Dynamic and Spatial
Dimension (CAPRI-DynaSpat)
Relevant spatial datasets for the disaggregation
of CAPRI-DynaSpat parameters Description - Use -
Constraints
Renate Köble and Adrian Leip
2
SPATIAL DATA SETS

• LAND COVER/LAND USE MAPS
• determine mainly the spatial resolution of the
  disaggregation
• LAND USE/COVER AREA FRAME STATISTICAL SURVEY
• can be used to create a decision matrix how to
  allocate the statistical agricultural activity
  data and model outputs to the land cover classes
• SPATIAL DATA ON ELEVATION, BIOGEOGRAPHICAL
  REGIONS AND SOIL
• deliver additional information to allocate
  statistical agricultural activity data and
  model outputs especially for complex land cover
  classes

3
CORINE LAND COVER/LAND USE 1990

• CORINE (Coordination of Information on the
  Environment) land cover mapping program was
  proposed in 1985 by the EU Commission to satisfy
  the need of precise and easy accessible
  information on land cover in Europe
• CLC describes land cover (and partly land use)
  according to a nomenclature of 44 classes
  organised hierarchically in 3 levels
• Elaborated based on the visual interpretation of
  satellite images and ancillary data (aerial
  photographs, topographic maps etc.)
• Acquisition period of satellite images 1985 to
  1995
• Smallest surface mapped 25 ha. Scale of the
  output product 1100 000
• The 100 m2 grid data set is available for the
  CAPRI-DynaSpat area of interest except Sweden
• For Switzerland a national land cover map is
  available with classes corresponding to Level II
  of the CORINE classification system

4
CORINE LAND COVER/LAND USE 2000

• An update of the CORINE Land cover database for
  the year 2000 is under processing
• The update will be more time consistent
  (satellite images from 2000 /-1year)
• Improvement of the geometric accuracy
• CORINE LC90 will be revised (land cover classes
  and geometry will be reviewed)
• Maps with land cover changes from 1990 to 2000
  will be produced
• Currently data is available for Ireland,
  Netherlands, Latvia, Luxembourg and Malta
• Data for Lithuania, Poland, Spain, Sweden, Italy
  might be available before summer
• the aim is to finish 80 of EU25 ( Bulgaria,
  Croatia, Romania) by the end of 2004

5
CORINE CLASSIFICATION
6
CORINE LC IN THE BONN AREA
CAPRI DYNASPAT KICK-OFF MEETING
7
PELCOM LAND COVER/LAND USE

• The Pan-European Land Cover Monitoring (PELCOM)
  project was carried out 1996-99. Funded as a
  shared cost action within FP4.
• The PELCOM land cover map distinguishes 14 land
  cover classes (4 agricultural classes)
• Classification is based on 1km resolution
  satellite images (NOAA AVHRR) and ancillary data
  as e.g. topographic information,
  rivers/lakes/coastlines
• Acquisition period of satellite images 1997
• Covers Europe and parts of Russia and the Middle
  East

8
CORINE/PELCOM LC CLASSIFICATION
CORINE
PELCOM
9
CORINE AND PELCOM LAND COVER
Pastures Complex cultivation pattern Land
princip. occ. by agric. sign areas of nat.
veg. Not irrigated arable land Forest Urban area
Grassland
Rhein-Sieg-Kreis
10
LAND COVER DATA SETS AVAILABLE FOR THE
CAPRI-DynaSpat AREA
11
COMPARISON OF CLC90 AND FARM STRUCTURE SURVEY
DATA
RECLASSIFIED STATISTICS
CLC90 11 agricultural classes, FSS 42 classes
Kayadjanian et al. (2001)
LANDCOVER MAP
12
THE POSSIBLE REASONS FOR THE DEVIATIONS ARE
MANYFOLD

• Data is related to different time spans (FSS
  1990, CLC 1985-95)
• Per definition CLC omits areas lt25 ha, therefore
  non irrigated arable land may be included to some
  extend also in other CLC classes as e.g. Complex
  cultivation patterns with significant area of
  natural vegetation or Grassland
• FSS classes can not be exactly regrouped in the
  CLC classes due to different classification
  systems (e.g. within irrigated land)
• Photo-interpretation inaccuracy for CLC
• Errors in the FSS

13
LUCAS SURVEY

• The Land Use/Cover Area Frame Statistical
  Survey (LUCAS) has been launched by Eurostat and
  DG Agri to
• obtain harmonised data (unbiased estimates) at EU
  15 level of the main Land Use / Cover areas and
  changes.
• evaluate the strengths and weaknesses of a point
  area frame survey as one of the pillars of the
  future Agriculture Statistical System (area frame
  means that the observation units are territorial
  subdivisions instead of agricultural holdings as
  in the Farm Structure Survey).

Decision N1445/2000/EC of the European
parliament and of the Council of the 22.05.2000
on the application of area-frame survey and
remote-sensing techniques to the agricultural
statistics for 1999 to 2003.
14
ORGANISATION OF THE LUCAS SURVEY

• Main land cover/use survey raster 18 km by 18 km
  with 10 subsampling Units
• Phase 1 field survey at 100000 observation
  points in EU15 (spring)
• Phase 2 interview with 5000 farmers to obtain
  additional technical or environmental information
  (autumn)
• The first survey has been carried out in 2001 (UK
  2002)
• 57 land cover classes are separated including 34
  agricultural classes
• High geometrical accuracy of the sampling
  locations (/- 3m)

Sampling design
Primary sampling units in NL
15
LUCAS SURVEY CLASSIFICATION
16
FINE SCALING CORINE LC CLASSES WITH LUCAS DATA

• Based on a study from J. Gallego (2002)

• Fine scaling in this case means estimating the
  proportion of other land cover classes within a
  given CORINE class as e.g. pastures
• To examine the possibility of fine scaling the
  CLC classes J. Gallego overlaid the CLC with the
  point observation of the LUCAS 2001 survey
• The operation produces a matrix with 56 columns
  (LUCAS land cover classes) and 44 rows (CLC) that
  allows to analyse the composition of other land
  cover classes within a specific CLC land cover
  class

17
FINE SCALING CLC 2000 WITH LUCAS DATA
LUCAS
Test for Ireland
CORINE
18
SPATIALISATION OF STATISTICALLAND USE WITHIN ONE
LAND COVER CLASS

• A study of the Geographical Information
  Management (G.I.M, 2002) group showed to possible
  value of using topographical (elevation, slope)
  and soil information to disaggregate CLC land
  cover classes with complex patterns into single
  classes .
• Example CLC class complex cultivation patterns
  contains 30 arable land, 40 pasture, 30 forest
  (based on CLC/LUCAS analysis). Roughly speaken
  arable land will be attributed to the best
  growing/farming conditions -gt good soils / low
  altitudes / flat terrain
• The G.I.M method will be reviewed
• Analysis if the assumptions can be improved by
  looking at relationships between LUCAS data
  soil topography

19
CHANGES IN AGRICULTURAL AREABASED ON CLC90 AND
CLC2000
94
20
LAND COVER CHANGES IN NL
Agricultural area to Artificial surfaces
Agricultural area to Forest seminatural areas
Amsterdam
Agricultural area to Wetlands
21
INFRASTRUCTURE FOR SPATIAL DATA IN EUROPE
(INSPIRE)

• With the INSPIRE initative, the European
  Commission intends to trigger the creation of a
  European Spatial Data Infrastructure (ESDI)
• The ESDI has to be set up in a way that will
  allow public users at European to local level to
  discover, access and acquire spatial data from a
  wide range of sources for a wide range of
  applications
• INSPIRE expert groups has been set up for several
  topics e.g.
• Reference data and metadata
• Data policy and legal issues
• Architecture and standards (reference system,
  projections, European reference grid system)