Evaluation of the Land Cover Features from Landsat TM over Saudi Arabia


DATA ANALYSIS AND RESULTS
The satellite used in this study was captured on 15/6/1998 (Fig. 2). A total of 42 training sample areas were selected in this analysis. For the satellite scene, six Landsat TM bands (except band 6) were used in the multispectral classification analysis using the classifiers mentioned earlier. The satellite image was classified into 3 classes, namely vegetation, soil and urban.

Three supervised classification methods were performed to the digital images (Maximum Likelihood, Minimum Distance-to-Mean, and Parallelepiped). Training sites were needed for supervised classification. Selection of training areas in this study was based on the colour image. A total of 200 samples were chosen randomly for the accuracy assessment. Many methods of accuracy assessment have been discussed in remote sensing literatures. Three measures of accuracy were tested in this study, namely overall accuracy, error matrix and Kappa coefficient. The most widely promoted and used accuracy measure, however, may be derived from a confusion or error matrix.

Kappa coefficient and overall accuracy values for the three classifications are shown in Table 1. The overall accuracy is expressed as a percentage of the test-pixels successfully assigned to the correct classes. Maximum Likelihood produced the highest degree of accuracy with overall accuracy of 91.2%, Minimum Distance-to-Mean gave overall classification accuracy of 75.2%, and Parallelepiped resulted in the overall classification accuracy of 41.0%. A classified image using Maximum Likelihood classifier is shown in Fig. 3.


Fig. 2 The study area imagery


Table 1: The overall classification accuracy and Kappa coefficient
Classification method Overall classification accuracy (%) Kappa coefficient
Maximum Likelihood 91.20 0.852
Minimum Distance-to-Mean 75.20 0.612
Parallelepiped 41.00 0.312



Fig. 3 The land cover changes map[Colour Code: Green = Vegetation, Blue = Land and Red = Urban]


CONCLUSION
From the three classified maps, Maximum Likelihood gives the best result for land cover mapping. The maximum likelihood supervised classifier produced the highest accuracy in this study. The classified map can be used to provide useful data for planning and management in theis area. The application of the Landsat TM imagery for land cover mapping produced reliable and accurate results.

ACKNOWLEDGEMENTS
We would like to thank the technical staff who participated in this project. Thanks are also extended to USM for support and encouragement

REFERENCES
  1. Adam, Y. H., Mohd Yusoff, A. R., Buse, I., Aman, M. S. and Redza, M., 2002, Proceeding of the International Symposium and Exhibition on Geoinformation 2002.
  2. Bruzzone, L. and Prieto, D. F., 2002, A partially unsupervised cascade classifier for the analysis of multitemporal remote-sensing images. Pattern Recognition Letters, 23, 1063–1071.
  3. Song, C., Woodcock, C. E., Seto, K. C., Lenney,M. P. and Macomber, S. A., 2001, Classification and Change Detection Using Landsat TM Data: When and How to Correct Atmospheric Effects? Remote Sensing Environment, 75, 230–244.
Page 2 of 2
Previous