The aim of this study was to develop an effective procedure for detecting land use/land cover (LU/LC) changes resulting from agricultural encroachment on eastern Mediterranean coastal dunes by using remote-sensing techniques. | Turk J Agric For 27 (2003) 135-144 © TÜB‹TAK A Remote Sensing Approach for Detecting Agricultural Encroachment on the Eastern Mediterranean Coastal Dunes of Turkey Süha BERBERO⁄LU* Department of Landscape Architecture, Faculty of Agriculture, Çukurova University, 01330, Adana - TURKEY Hakan ALPHAN Department of Landscape Architecture, Faculty of Agriculture, Çukurova University, 01330, Adana - TURKEY Kemal Tuluhan YILMAZ Department of Landscape Architecture, Faculty of Agriculture, Çukurova University, 01330, Adana - TURKEY Received: Abstract: The aim of this study was to develop an effective procedure for detecting land use/land cover (LU/LC) changes resulting from agricultural encroachment on eastern Mediterranean coastal dunes by using remote-sensing techniques. Historic LU/LC information was extracted from aerial photos taken in 1976 and IKONOS imagery was acquired in 2002 to determine the current LU/LC pattern. The remotely sensed aerial and satellite data were classified by integrating spectral information with measures of texture, in the form of statistics derived from the variance, co-occurrence matrix and variogram. The performance of these classification approaches was evaluated in terms of error matrices. The accuracy of the classification was greater when using the variogram texture measure and spectral data together than when using spectral data alone or incorporating with co-occurrence matrix texture for both image classifications. Co-occurrence texture measures did not result in a significant increase in accuracy with the maximum likelihood classifiers in the classification of IKONOS imagery. The addition of variogram texture information in particular, with variogram lags of 1, 2 and 3 pixels, increased the overall classification accuracy by . The principal conclusion of this paper is that the accuracy of agricultural land cover and semi-natural vegetation classification can be maximised by incorporating textural information .
