عنوان مجله
|
Journal of the Indian Society of Remote Sensing
|
کلیدواژهها
|
Classification , Gully , Lighvan watershed , Object-oriented algorithms , Segmentation
|
چکیده
|
Gully erosion and its complex and variable characteristics have led to challenging and many problems for creation of
erosion maps with the most commonly used methods of satellite imagery analysis. The fuzzy algorithm-based object-based
image analysis (OBIA) method is one of the most effective methods for classifying satellite imagery, which aims to
integrate the spectral satellite imagery and provides the necessary facilities for the use of environmental and spatial
information as well as a geometric feature of land surface phenomena. The purpose of this study was to apply and evaluate
the implementation of different classification algorithms and fuzzy functions membership detecting and mapping of gully
erosion. For this purpose, satellite images of Sentinel-2 were employed and accordingly the soil erosion map of Lighvan
watershed was developed within the semi-automated approach by applying fuzzy-OBIA features and techniques. The
obtained results from the accuracy assessment of the methods indicated that the accuracy of gully maps obtained by fuzzy-
OBIA features including: length/width, asymmetry, density, and shape index algorithms with the respective kappa coefficient
of 0.78, 0.91, 0.85, and 0.89 is higher than other algorithms. Also, the results obtained from the study of the degree
of membership of fuzzy functions indicate the high role of this factor in the accuracy of the results, so that the highest
classification accuracy (kappa coefficient greater than 0.91) was related to the length/width algorithm, in which the fuzzy
functions B, D, H and L with the highest degree of membership (0.685, 1, 0.972 and 1, respectively) were employed.
Results of current research are great of important for decision makers and authorities by means of providing detailed
information regarding the soil erosion in the study area as well as GIS sciences society for applying and introducing the
most efficient methods and techniques.
|