Ammar Rafiei Emam
M. Sc student of Arid Lands Reclamation Management,
Natural Resources Faculty, Tehran University, Tehran, Iran.
Tel: +98 911 2499262, Fax: +98 21 8966760
Email: [email protected]
S. Kazem Alavipanah
Assistant professor, Iran Desert Research Center
University of Tehran, Tehran, Iran.
The remote sensing monitoring and also GIS techniques are very useful for land resource management. The satellite imageries spatially TM multi-spectrum data and also GIS techniques help to studies of land use.
In this research, for survey of land use in the varamin plain, one of the great plain in Tehran province, Iran, were used from Landsat TM images recorded on May 18th, 1998.
After investigation of images quality, Suitability of bands selected with help of OIF index and also correlation matrix between bands for supervised classification, and classification was done with maximum likelihood algorithm and seven classes (farmlands, follow lands, range lands, barrenlless, salt crusts, wet lands and urban) have separated.
The accuracy of classified image were assessed by training samples, the results show that the overall accuracy is low and omission and commission errors in some of classes are high, therefore for increase of accuracy used from GIS.
The GIS was applied for correcting the wrong pixels of some classes. To correct the urban, wetlands and follow lands classes that had high omission and commission errors the GIS technique were applied. This classification that used from GIS and R.S. was called Hybrid classification.
At the end accuracy of hybrid classification were assessed again.
The results showed that GIS techniques help to classification and increase of overall accuracy.