A digital approach to the separation of parabolic sand dune areas from interdunal areas using Landsat MSS data

An appropriate methodology for the digital discrimination between sand dune area and interdunal areas has been elaborated. Dunes are easy to distinguish visually on Landsat MSS imagery due to the sharp contrast between the sunlit and the shadowed side. In a digital approach, texture measures have to be incorporated. Two first order texture transformations were applied, preceeded by several enhancement procedures to emphasize the contrast in the image. The texture images were classified into two classes: ʹduneʹ and ʹinterdunal areaʹ by means of an unsupervised classification. The evaluation of the procedures was done by mutual comparison of the Kappa coefficient of the different classifications. The test area on which the methodology has been worked out is situated in the Thar Desert, Rajasthan, India.