[
t
anbul 2004
47.90 «
2.087050
1.992803
1.438374
0.000645
1024.343038
-525.664732
1141160
1141158
m
"om n ow ow o" m
and Linear
another and
elation was
ap and the
n results are
lihood-linear
discriminant
d classifiers
'se results is
lihood and
efficient than
t classifiers
parellelpiped
elpiped map
surface. It is
on algorithm
PTS:
= 57.72 +
= 2.087050
= 2.111239
= 1.272756
0.000616
"onu
'
e
-
co
o
ne
=
= 1141160
= 1141158
ind. Minimum
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
1933054 Q05578040X — re
Regression Parameters:
X axis: fisher
Y &xis: mindist
Coeff. of Det. = 27.73 %
Std. Dev. of X = 1.992803
Std. Dev. of Y = 2.111293
S.E. of Estinate = 1.794294
Std. Error of Beta = 0.000943
t Stst for r or Beta 661.656027
t Stat for Beta <> 1 = -524 351238
Saxple Size (nj = 1141160
Apparent df = 1141158
high frequency
Figure 4. Regression of Linear discriminant and Minimum
distances classifiers.
| Y= 1165727+ 0.763452X t= 0762671
Regression Parameters:
X axis: maxlike
Y axis: piped
el Coeft. of Det. = $8.20 1
Std. Dev. of X = 2.087050
| Std. Dev. of Y = 2.088639
154- S.E. of Estinate = 1.350412
i Std. Error of Beta = 0.000606
ly t Stat for r or Beta = 1260.436523
| t Stat for Beta <> ] = -350.534486
| Sauple Size (n) = 1141160
3 Apparent df = 1141158
high frequency
l'igure 5. Regression of Maximum likelihood and Parellelpiped
classifiers.
4. CONCLUSIONS
For more effective use of the satellite remote sensing, landuse
managers should be aware of the limitations and advantages of
satellite data and should chose from their avaible landuse
mapping options accordingly. Remote sensing is especially
proper for initial reconnaissance mapping and continued
monitoring of landuse over large areas. In this context,
techniques for improving the classification of landuse with
satellite remote sensing data include the use of appropriate
digital data. In order to achieve this task, selection of the most
proper satellite image, band combination, and the classifier are
Very important. Additionaly, the image processing is important
and different stages of it such as filtering of bands and principal
Component analysis should be applied before evaluation. All
these points were applied to this study and it has been seen that
Maximum likelihood classifier was the most suitable
1095
classification method for landuse mapping purpose. Minimum
distances classifeir was also determined as suitable as the
maximum likelihood classifier.
4.1 References
Butera, M.K., 1983, Remote sensing of wetlands, [EEE
Transactions on Geoscience and Remote Sensing GE-21,pp.
383-392
Campbell, 2002, Introduction to Remote Sensing,
CORINE Land Cover Technical Guide, European Commission,
Luxemburg, pp. 21-53.
Dean, A.M and Smith, G.M., 2003, An evaluation of per —
parcel land cover mapping using maximum likelihood class
probabilities, /nternational Journal of Remote Sensing, 24 (14),
pp. 2905-2920
Ernst, C.L. and Hoffer, R.M., 1979, Digital processing of
remotely sensed data for mapping wetland communities. LARS
Technical Report 122079. Laboratory -for Applications of
Remote Sensing. Purdue University, West Lafayette, 119 pp.
Liu, X.H., Skidmore,A.K. and Oosten, V.H., 2002, Integration
of classification methods for improvement of land-cover map
accuracy, ISPRS Journal of Photogrammetry&Remote Sensing,
56, pp 257-268.
Lo, C.P. and Watson, L.J., 1998, The influence of geographic
sampling methods on vegetation map accuracy evaluation in a
swampy environment. Photogrammetric Engineering and
Remote Sensing 64:1189-1200.
Ozesmi, S.L and Bauer, M., 2002, Satellite Remote Sensing of
Wetlands, Wetlands Ecology and Management, 10, pp. 381-402
Richards, J.A., 1995, Remote Sensing Digital Image
Analysis: An Introduction, Springer-Verlag, pp. 265-290.
IDRISI Klimanjaro, 2004, Guide to GIS and Image Processing
Volume 2, Idrisi Production, Clark Labs,USA, pp.57-82
4.2 Acknowledgments
The authors would like to thank to Dokuz Eylul University
Scientific Research Fund for providing financal support for this
project (Project no: 02.K B.FEN.052).