Overview and Brief History - Sampling Spatial Units for Agricultural Surveys - page 43

but also all other types of territory. LUCAS has a double nomenclature: each point
has a land cover code (57 classes) and a land use code (14 classes).
In 2006, the design was modified to a stratified two-phase sampling of
unclustered points (Jacques and Gallego 2005 ). The stratification was performed
by photo-interpretation of a 2-km grid of points in EU25. 4 The points placed on
small islands were not considered in the sample (i.e., Baleares, Azores, the Canary
Islands, Cyprus, Malta, and the Greek islands except Crete). A ground survey was
executed on a sub-sample of 169,000 points covering 11 countries, over about 70 %
of the area of EU25.
The sampling design was again modified for the LUCAS 2009 to give more
importance to environmental and agri-environmental parameters. The 2009 sam-
pling strategy aimed at providing precise estimates at the NUTS 5 1 level. Following
these requirements, a sample at EU NUTS 2 level has been selected. The regions
(i.e., NUTS 2) were divided into two groups according to their total area. Group A is
composed of NUTS 2 regions with a total area less than or equal to 500-km 2 , while
Group B contains NUTS 2 regions with a total area above 500-km 2 . Additionally,
Group B was partitioned into two parts: B1 contains regions belonging to the
11 countries already observed in 2006, and B2 contains regions from the other
countries. For Group B1, some auxiliary information was available from the
previous survey.
A different sampling strategy was chosen for the different groups. For Group A,
the precision was not fixed, and the units were allocated to the strata in proportion to
their size. For regions in Group B1, auxiliary information was used, and an optimal
allocation (see Sect. 8.3 ) sampling scheme was formulated by exploiting a set of
land cover classes. For Group B2, no information was available from the 2006
LUCAS survey. In this case, the auxiliary information was the land cover and land
use data collected by the Corine Land Cover (CLC) program of the European
Environment Agency (EEA 2007 ). The CLC classes were divided into 12 new
classes. Based on this information, dissimilarity indexes were calculated among
regions belonging to Groups B1 and B2 according to the city block distance

X

d ¼ j x cr x cr 0 j; ð 2 : 2 Þ

c

where x cr and x cr' are the areas of CLC group c in region r and r' , respectively. The
choice of maximizing the distance between each sample point (in both the same and
different strata) was decided according to the 2006 survey, because it was shown to
be efficient (Jacques and Gallego 2005 ). Conversely, points sampled in different

4 The EU25 (1 May 2004-31 December 2006) contained the EU15 countries plus Cyprus, Czech
Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Slovakia, and Slovenia.
5 The NUTS is a standard geocode classification for the subdivisions of countries for statistical
purposes. The classification was developed by the EU. For each EU member country, there is a
hierarchy of three NUTS levels.