Overview and Brief History - Sampling Spatial Units for Agricultural Surveys - page 46

The first survey on the use of agricultural land dates back to 1946. The design of
that survey was based on the land surface from cadastral maps. In 1962, aerial
photography was introduced, not as a support for the survey but as a tool for
updating cadastral maps. From 1969, this design approach was generalized to all
French departments.
The first TER-UTI was carried out in 1982. It was renewed in 1990 and 1991 to
correct certain biases introduced during sample selection in 1982. Continuous time
series are available for the years 1982-1990 and 1992-2004 for 550,000 points,
except in 2004 when the sample was reduced to 155,000 points.
In 2005 the survey design was re-defined. Two main developments led the
researchers to modify the survey. First, technical progress in the digitization of
cartographic and geo-referencing points encouraged the re-definition of the sam-
pling points TER-UTI, which were previously determined manually using aerial
photographs. Second, the definition of the LUCAS survey by EUROSTAT (based
on the same methodological principles as TER-UTI) encouraged a deep revision to
produce consistent nomenclature, observation methods, and sample design. For
these reasons, the French survey is now denoted as TER-UTI LUCAS.
The main objectives of TER-UTI LUCAS are:
• To determine, each year, the different classes of land cover and land use for the
whole territory (agricultural, environmental, and urbanized) at different geo-
graphical levels (national, regional, and departmental).
• To monitor and to quantify the land cover and land use, and landscape structure
changes.
• To provide a data source for other studies (agri-environmental indicators).
• To increase the sample (number of segments or number of points per segment).
The population under investigation extends to almost the entire national terri-
tory. The only restrictions are Guyana and Mayotte, which are not suited to this
method because of their small sizes.
TER-UTI LUCAS uses a non-stratified two-stage sampling scheme, with points
grouped in PSU. The PSUs are the segments, generally identified by a square area
that varies from 1.5 km 600 m to 1.5 km 1.5 km. The second stage units are the
points, which have a 3-m radius circle centered on the point (basic observation
window) in the general case, or a 40-m radius circle (extended observation window)
in the case of heterogeneous land cover. Points are spaced 300 m inside a segment.
The two-stage sampling scheme is a compromise between the cost of the survey and
the desired precision of the results. It presents almost the same precision as a
one-stage sampling but is less expensive.
Data from this survey are used in a number of ways (Vidal and Marquer 2002 ).
In particular, the data set has been used for the calculation of three indicators: a
predominant cover indicator, a spatial organization indicator, and a temporal trend
indicator. To compute these indicators, the points are grouped into three categories:
natural, agricultural, and non-natural (i.e., artificial). Natural areas comprise rock
and water, moorlands, grazing land, alpine pastures, deciduous forests, and conif-
erous forests. Agricultural zones include hedges, scattered trees, poplar stands,