Document Type : Research Paper

Authors

• Mehdi Tazeh ¹
• Maryam Asadi ²
• Rouhollah Taghizadeh Mehrjerdi ¹
• Saeedeh Kalantari ¹
• Majid Sadeghinia ¹

¹ Assistant Professor, Faculty of Natural Resources, Ardakan University, Ardakan, Iran

² D. Student of Watershed Management, Department of Reclamation of Arid and Mountainous Regions, Faculty of Natural Resources, University of Tehran, Karaj, Iran

Abstract

Geomorphological map is one of the main information layers in natural resources studies. So far, various methods have been proposed for the classification and separation of various units and Geomorphological types, most of which are based on qualitative and descriptive information. In this study, the ability of geomorphometry parameters in separation of mountains from pediment and also separation of different types of pediments was investigated. First, ground truth map was prepared using visual interpretation of satellite data and topographic maps. Then the 1000-point sampling grid was designed   randomly. Parameters including profile curvature, plan curvature, tangential curvature, cross-sectional curvature, longitudinal curvature, and general curvature were prepared from digital elevation model in the GIS software. Then, their values were extracted at all points of the sampling network. Then, artificial neural network with structure of 13_6_ 4 was used to separate the units. The results showed that the erosion pediment could be separated from epandage using artificial neural network; however, the separation of epandage pediment from covered pediment was not well. For this purpose, to improve network performance, the digital value of Landsat 7 data was added to the previous values. The resolution accuracy of mountain, erosion pediment, epandage pediment, and covered pediment was calculated to be 90, 79, 80, and 76%, respectively.

Keywords

• Geomorphometry parameters
• Artificial Neural Network
• satellite images
• semi-automated methods