Application of TGSI - Albedo feature space model in assessing of desertification status in the center of Khuzestan province

Abstract

Nowadays, desertification and land degradation are among the most important environmental chalenges caused by various factors, including climate variation and human activities, especially in arid and semi-arid regions. Therefore, understanding of the desertification status is essential for future management and planning. In this regard, using remote sensing indicators to prepare basic desertification maps and its monitoring can be efficient. In this study, Topsoil grain size and albedo indices were used as indicators to identify desertification in the center of Khuzestan province. After constructing the above-mentioned indices using Landsat ETM+ sensor image, the values of 411 randomly selected samples on the images were used to construct the Albedo-TGSI feature space model. The correlation between the variables was 0.83. The DDI desertification degree equation is then obtained based on the slope of the fitted line. In the next step, by applying natural break classification on the DDI index, different levels of desertification and the break values ​​were obtained for random samples. These break values ​​are then applied to the whole study area, and finally, the 2018 desertification status map was obtained. Results indicated that about 70% of the study area fell under the severe and high desertification intensities, whereas 18.3%, 8.3% and 4.1% fell under the medium, low and none desertification grades respectively. The accuracy of the produced map with a kappa of 92.1% and an overall accuracy of 94.3% showed that the feature space model is a useful and robust tool for extracting desertification degrees in barren lands or areas with extremely low vegetation coverage.

Keywords

• desertification
• classification
• remote sensing
• spectral reflectance