Research Methodology - Historical Land Use/Land Cover Classification Using Remote Sensing - page 134

study area for classification of the irrigated winter and summer crops during the
previously mentioned time durations except for in 1975. The remotely sensed data
(LANDSAT-MSS) available in that year had a low spatial resolution of 60 m, and
for this reason, it was impossible to produce detailed maps of land uses, specifi-
cally in those areas included under wide classes (e.g., both wheat and barley are
detailed classes that lie under the heading ''wide general class'', namely, the
agrarian lands).
The total cultivated area of the Arraed project was c. 21,000 ha. However, due
to bad land reclamation procedures, salinization had resulted in large areas of the
project lands. In 2005, the arable land mass was only 2,433 ha.
5.10 Crops Classification
Classification of agricultural crops using remote sensing data requires in general,
knowledge about crop phenology, climate of the exacting growing season and
ground reference information about specific agricultural practices in the drainage
basin. ''The development of a regional-scale crop mapping methodology is chal-
lenging because it requires remotely sensed data that have large geographic cov-
erage, high temporal resolution, adequate spatial resolution relative to the typical
field size, and minimal cost''. Remotely sensed data from customary sources such
as the LANDSAT (TM and ETM+) and (AVHRR) proved the usefulness for the
classification of LULC-features. Supervised classification is the most frequently
used classification method in agricultural areas (Van Niel and McVicar 2000 ).
MSS data are used to set maps concerning the expansion of the agricultural
lands and to distinguish them from constructional lands, for example. However,
one drawback is that these data are unable to set thematic maps which view the
more detailed crops' expansion. Of course, there are always exceptions; for
instance, the agricultural cultivated fields planted with different crops are to some
extent considered wide spaces, which enables the MSS-data to distinguish them.
Yet, this condition was not verified in the ERB, which was characterized by having
small agricultural fields, especially those located outside of the borders of the
governmental irrigated agrarian projects. These areas were also organized fol-
lowing the agricultural crop rotation policy.
In the third part of this study, carefully timed remotely sensed data were used to
map the location and extent of irrigated winter and summer crops for the years
1987, 2005 and 2007.
The commonly implemented crop classification approaches included: unsu-
pervised classification; supervised classification; and decision tree classifier. In the
cases where there was less information for a study area, only the characteristics of
the image (also, statistical records and the detailed schemes, especially for 1987)
were used.