In: Wagner W„ Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B
568
ANALYSIS OF RAPIDEYE IMAGERY FOR ANNUAL LANDCOVER MAPPING AS AN
AID TO EUROPEAN UNION (EU) COMMON AGRICULTURAL POLICY
Brooke Tapsall, Pavel Milenov and Kadim Taçdemir*
Monitoring Agricultural Resources Unit, Institute for the Protection and Security of the Citizen,
European Commission Joint Research Centre
Via E. Fermi 2749, 21027, Ispra (VA), Italy
{brooke.tapsall, pavel.milenov, kadim.tasdemir}@jrc.ec.europa.eu
http : //mar s. j rc. ec. europa. e u/
KEY WORDS: Agriculture, GIS, Land Cover, Change Detection, Classification, Farming, Modelling, Monitoring
ABSTRACT:
The Common Agricultural Policy (CAP) of the European Union (EU) was established to maintain balance between farming industries
and the environment as well as to provide economic sustainability in rural areas. EU Regulations for agricultural and rural development,
adopted by countries upon their admission to the EU, allow payments to farmers for each eligible hectare of agricultural land (CAP
reform), under different payment schemes. Remote sensing data is currently used as an efficient tool in determining areas potentially
eligible for payments, through land cover identification and mapping. Launched in August 2008, RapidEye consists of five constellation
multispectral sensors with a ground sampling distance (GSD) of 6.5m and a daily overpass. The satellite has a predicted lifespan of
7 years and with the target application of the sensor being agriculture; contains a high potential for the application of agricultural
monitoring, necessary to some new Member States, such as Bulgaria and Romania. Analysis of RapidEye imagery, combined with
local ancillary data over pre-selected test zones lead to determination and classification of land cover features which have potential or
no potential to be eligible under the Single Area Payment Scheme (SAPS). This classification was completed using object oriented
analysis and was run concurrently alongside a pixel based (self-organizing maps) analysis for comparison.
1 INTRODUCTION
The European Union (EU) established the Common Agricultural
Policy (CAP) to support the agricultural sector in Europe, assess
its impact to the environment and ensure economic sustainability
in rural areas. One of the principal payment schemes under the
CAP is the Single Area Payment Scheme (SAPS) which regulates
payment of uniform amounts per eligible hectare of agricultural
land. For most EU member states applying SAPS, the agricul
tural area eligible for payments is the utilised agricultural area,
maintained in good agricultural condition (GAC) at a given ref
erence date. As a consequence, the land which can be declared
by the farmers and is the subject of the administrative and con
trol processes that manage the CAP payments is limited to the
historical extent, fixed at the reference date. Two exceptions are
Bulgaria and Romania where the requirement for the reference
year was omitted in their accession treaties. As a result, for these
countries any utilised agricultural area, maintained in good agri
cultural condition at the time of the farmer declaration, regardless
of its past status, can be considered eligible for payment. This
creates a particular challenge for land management in the years
following the EU accession, as agricultural land eligible for pay
ment should be assessed on annual basis.
is to clarify the concept of what is ‘good agricultural condition’
(GAC) in the national context, as there is no common legal defi
nition of GAC at EU level. The proposed methodology envisages
remotely sensed imagery, as an efficient source of up-to-date in
formation, to detect and quantify (for the entire country) the agri
culture land, that may represent eligible area, through monitoring
of land cover dynamics. The recently launched constellation of
RapidEye satellites was considered particularly suitable for this
study, as the satellites were designed to be used mainly for moni
toring of agricultural and natural resources at relatively large car
tographic scale. The methodology was based on multi-temporal
analysis of RapidEye time-series.
In order to detect eligible agricultural land and estimate their im
pact at reference parcel level, two different approaches were con
sidered: i) object oriented classification techniques (Gamanya et
al., 2007, Mathieul and Aryal, 2005) based on red edge normal
ized difference vegetation index (Wu et al., 2009, Gitelson et
al., 1996) and ii) automated clustering of self-organizing maps
(Taçdemir and Milenov, 2010). The proposed methodology was
tested using zones selected according to the variability of land
cover features across the country, which potentially represent el
igible land (Milenova et al., 2001).
The objective of this study is to investigate and develop an opera
tionally efficient methodology for annual monitoring and assess
ment of land eligible for subsidy payments under SAPS in Bul
garia. In order to ensure a correct assessment of the agricultural
land suitable for SAPS payments, a necessary preliminary step
*This study is a joint research between the GeoCAP Action of the
Monitoring Agricultural Resources (MARS) Unit at the Institute for the
Protection and Security of the Citizen in Joint Research Centre of Eu
ropean Commission, the Bulgarian Ministry of Agriculture and Food,
ASDE/RESAC, Bulgaria and RapidEye AG, Germany. Figures in this
paper are in color. Request color reprint from the authors.
The outline of the paper is as follows: Section 2 introduces the
concept of Good Agricultural Condition (GAC) elaborating on a
proposal for its legal definition; Sections 3 and 4 briefly overview
the test areas of the study and RapidEye sensor specifications;
Section 5 describes proposed methodology for detection and quan
tification of the GAC/non-GAC land cover types and features,
using object oriented approach; Section 6 briefly presents the
first results of the study and the ongoing validation, Section 7
provides results from a concurrent testing using Self-Organizing
Maps, as an alternative of the object oriented approach; Section 8
concludes the paper.