bul 2004
:lassified
). Kappa
ation of
1 in error
> error of
nty basis
ural land
ld crops
/ineyards
nty.
rops
(%) in
total
agric.
area
(84 % of
vered by
io areas
/ ineyards
est of the
ere given
| between
all
y (%)
10
39
28
10
38
n used to
was only
ysis were
Id supply
vioreover,
ive to get
ond crop
udy area.
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
On the other hand, main agricultural crops of pistachio, olives
and vineyards showed similar reflectance because of
background soil reflectance. These crops are grown without
irrigation with sparse stands. For this reason, bare ground
affects the reflectance of those three crops similarly. Images of
different dates (spring time, summer time) didn't help to
differentiate between pistachio, vineyards and evergreen olive
groves in the classification process. To overcome this problem,
as many as ground truth data were used and the number of
training signatures in supervised classification algorithm was
increased and extended across the image. Photo interpretation
was also used to increase accuracy of digital classification.
Wrongly classified pixels whose real class type are known are
manually edited to include in related class.
4. Conclusion
Agricultural management highly depends on knowledge land
use information. There are many agricultural surveys and land
use maps prepared and completed in the past. Undoubtedly all
of them are important sources giving the idea about agricultural
land but they are far from giving up-to-date information for
effective management.
Remotely sensed data has filled a gap about the needs of up-to-
date land use information. Basically, the use of this kind of data
is getting to spread in Turkey. This study is one of the
pioneering projects that use remote sensing techniques
effectively. Although
This study showed that remotely sensed data, especially
LANDSAT-TM images, are highly suitable to determine the
current land cover and land use characteristics in a cheap, fast,
and accurate way.
Applied supervised classification with maximum likelihood
parametric rule and LANDSAT-TM images were found suit-
able to display the land cover and land use characteristics of the
area. Therefore, decision makers who are studying on agricul-
ture can utilize this kinds of information satisfactorily. This
study showed how a complete agricultural spatial database
could be a good starting point for many planning applications.
163
References
ERDAS, 1999, ERDAS Field Guide Fifth Edition, ERDAS
Inc., Atlanta, Georgia.
Jensen , J. R. 1996. Introductory Digital Image Processing.
Prentice Hall, Upper Seddle River, New Jersey, pp: 225-231
Verbyla, D. 1995. Satellite Remote Sensing of Natural Re-
sources, Lewis Publishers Boca Raton, Florida, pp: 77-95.
Weber, R.M. and Dunno, G. A. 2001. Riparian Vegetation
Mapping and Image Processing T
echniques Hopi Indian Reservation, Arizona. Photogrammetric
Engineering and Remote Sensing, 67(2): 179-186
Zhuang, X., Engel, B. A., Xiong, X., Johannsen, C. J. 1995,
Analysis of Classification Results of Remotely Sensed Data and
Evaluation of Classification Algorithm. Photogrammetric Engi-
neering and Remote Sensing, 61(4): 427-428.
